Pivotal GemFire Native Client Supported Configurations

Pivotal GemFire Native Client Supported Configurations
Product Documentation
Pivotal GemFire® Native Client
Version 7.0
User's Guide
Rev: 12
© 2015 Pivotal Software, Inc.
Copyright
Pivotal GemFire Native Client
Notice
Copyright
Copyright © 2014 Pivotal Software, Inc. All rights reserved.
Pivotal Software, Inc. believes the information in this publication
is accurate as of its publication date. The information is subject to
change without notice. THE INFORMATION IN THIS PUBLICATION
IS PROVIDED "AS IS." PIVOTAL SOFTWARE, INC. ("Pivotal")
MAKES NO REPRESENTATIONS OR WARRANTIES OF ANY KIND
WITH RESPECT TO THE INFORMATION IN THIS PUBLICATION,
AND SPECIFICALLY DISCLAIMS IMPLIED WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
Use, copying, and distribution of any Pivotal software described in this
publication requires an applicable software license.
All trademarks used herein are the property of Pivotal or their
respective owners.
2
Copyright
Pivotal GemFire Native Client
3
Contents
Pivotal GemFire Native Client
Contents
Preface: Contacting Pivotal........................................................................x
Current Pivotal Customers.................................................................................................................. x
Part I: Pivotal GemFire Native Client User's Guide................................12
Part II: Supported Configurations and System Requirements.............. 14
Pivotal GemFire Supported Configurations.......................................................................................15
Host Machine Requirements.................................................................................................. 17
Dependencies on Linux RPM Packages................................................................................17
Running Pivotal GemFire in Pure Java Mode........................................................................18
Pivotal GemFire Tools Supported Configurations.............................................................................18
GemFire Pulse System Requirements................................................................................... 18
DataBrowser System Requirements...................................................................................... 18
VSD System Requirements....................................................................................................19
Pivotal GemFire Native Client Supported Configurations................................................................. 20
Pivotal GemFire Modules Supported Configurations........................................................................ 22
Part III: Getting Started with Pivotal GemFire Native Client.................. 24
About the Native Client..................................................................................................................... 25
Installing the Native Client................................................................................................................ 25
Installing on Linux and Solaris............................................................................................... 26
Installing on Windows.............................................................................................................26
Uninstalling the Native Client................................................................................................. 28
Running Native Client Applications................................................................................................... 29
Developing C++ Programs on Linux...................................................................................... 29
Developing C++ Programs on Solaris....................................................................................30
Developing C++ Programs on Windows................................................................................ 31
QuickStart Examples and Guide.......................................................................................................33
Running the QuickStart Examples......................................................................................... 33
Accessing the QuickStart Guide............................................................................................ 33
Part IV: Setting System Properties..........................................................34
Configuring the Native Client and Cache Server..............................................................................35
Native Client Configuration.....................................................................................................35
Cache Server Configuration................................................................................................... 36
Attribute Definition Priority...................................................................................................... 36
Search Path for Multiple gfcpp.properties Files..................................................................... 37
Overriding gfcpp.properties Settings...................................................................................... 37
Defining Properties Programmatically.................................................................................... 37
Attributes in gfcpp.properties.............................................................................................................37
gfcpp.properties Example File...........................................................................................................42
Using the Default Sample File............................................................................................... 43
Part V: Configuring the Native Client Cache.......................................... 46
4
Contents
Pivotal GemFire Native Client
Caches............................................................................................................................................... 47
About the Native Client Cache...............................................................................................47
Cache APIs.............................................................................................................................47
Local, Remote, and Distributed Caches.................................................................................48
Creating and Accessing a Cache...........................................................................................48
Closing the Cache.................................................................................................................. 49
Cache Initialization File (cache.xml)..................................................................................................49
Cache Initialization File Basics...............................................................................................49
Example cache.xml File..........................................................................................................50
Native Client Cache XML DTD.............................................................................................. 51
Regions.............................................................................................................................................. 55
Declarative Region Creation...................................................................................................55
Programmatic Region Creation.............................................................................................. 56
Invalidating and Destroying Regions......................................................................................56
Region Access........................................................................................................................57
Getting the Region Size......................................................................................................... 57
Region Entries................................................................................................................................... 57
Entry Distribution Requirements.............................................................................................58
Registering Interest for Entries...............................................................................................58
Using serverKeys to Retrieve a Set of Region Keys............................................................. 60
Adding Entries to the Cache.................................................................................................. 60
Updating Entries..................................................................................................................... 61
Accessing Entries................................................................................................................... 62
Invalidating or Destroying Cached Entries............................................................................. 63
Notification for Operations...................................................................................................... 63
Region Consistency...........................................................................................................................63
Region Attributes............................................................................................................................... 64
Specifying Region Attributes.................................................................................................. 64
RegionShortcuts......................................................................................................................64
Mutable and Immutable Region Attributes............................................................................. 65
CachingEnabled...................................................................................................................... 66
InitialCapacity.......................................................................................................................... 67
LoadFactor.............................................................................................................................. 67
ConcurrencyLevel................................................................................................................... 67
ConcurrencyChecksEnabled...................................................................................................67
LruEntriesLimit........................................................................................................................ 68
DiskPolicy................................................................................................................................68
PersistenceManager............................................................................................................... 69
Specifying Application Plug-In Attributes................................................................................74
Specifying Expiration Attributes..............................................................................................79
Cache Management.......................................................................................................................... 80
Client-to-Server Connection Process..................................................................................... 80
Controlling Cache Size........................................................................................................... 81
Managing the Lifetime of a Cached Object........................................................................... 82
Using Thread Safety in Cache Management......................................................................... 83
Troubleshooting...................................................................................................................... 83
Part VI: Pivotal GemFire Native Client C++ API......................................86
About the Native Client C++ API...................................................................................................... 87
Creating a Cache.............................................................................................................................. 88
Creating a Proxy Client-Side Region................................................................................................ 89
Adding an Entry to the Cache.......................................................................................................... 90
Accessing an Entry........................................................................................................................... 91
Serializing Data................................................................................................................................. 91
Region Data Requiring Serialization...................................................................................... 91
5
Contents
Pivotal GemFire Native Client
Data Serialization Options...................................................................................................... 92
Serializing Data with GemFire PDX Serialization...................................................................93
Serialize Your Domain Objects with PdxSerializer and PdxWrapper.......................... 93
Serialize Using the GemFire PdxSerializable Class....................................................95
Programming Your Application to Use PdxInstances..................................................97
Configuring PDX to Ignore Unread Fields During Deserialization............................... 98
Using PdxInstanceFactory to Create PdxInstances.................................................... 98
Using C++ Enum Type with PDX Serialization......................................................... 100
Using PDX Serialization with Delta Propagation....................................................... 102
Serializing Data with the GemFire Serializable Interface..................................................... 102
Serializing Object Graphs.....................................................................................................107
Serializing and Accessing Data as a Blob........................................................................... 107
Implementing User-Defined Objects in Java Clients.......................................................................107
Using a Custom Class.................................................................................................................... 108
Creating New Statistics................................................................................................................... 109
Part VII: Pivotal GemFire Native Client C# .NET API............................112
About the Native Client C# .NET API.............................................................................................113
C# .NET Naming and Usage Conventions.......................................................................... 114
Primary APIs in GemStone::GemFire::Cache::Generic........................................................114
Cache APIs................................................................................................................ 114
Region and Entry APIs.............................................................................................. 114
Data Serialization APIs.............................................................................................. 115
Event Handling APIs..................................................................................................115
Property Collections and Logging APIs.....................................................................116
C++ Class to .NET Class Mappings............................................................................................... 116
Java to .NET Type Mapping Table................................................................................................. 118
Object Lifetimes...............................................................................................................................120
.NET Application Domains.............................................................................................................. 121
Problem Scenarios............................................................................................................... 121
Creating a Cache............................................................................................................................ 122
Creating a Region........................................................................................................................... 122
Adding an Entry to the Cache........................................................................................................ 123
Accessing an Entry......................................................................................................................... 124
Data Serialization............................................................................................................................ 124
Data Serialization Options.................................................................................................... 125
Serialize with GemFire PDX Serialization............................................................................ 126
GemFire PDX Serialization Features........................................................................ 126
Serialize Using the GemFire Pdx Autoserializer....................................................... 127
Extend the PDX Autoserializer.................................................................................. 128
Serialize Your Domain Objects with IPdxSerializer................................................... 129
Serialize Using the GemFire IPdxSerializable Interface............................................ 129
Program Your Application to Use IPdxInstance........................................................ 131
Use the IPdxInstanceFactory to Create IPdxInstances............................................. 132
Map .NET Domain Type Names to PDX Type Names with IPdxTypeMapper...........134
Serialize with the GemFire IGFSerializable Interface.......................................................... 135
Generic and Custom Serializable Types..................................................................135
How Serialization Works with IGFSerializable......................................................... 135
Implement the IGFSerializable Interface................................................................... 135
Register the Type...................................................................................................... 137
Using a Custom Class With IGFSerializable........................................................................138
Application Callbacks................................................................................................. 140
A Simple C# Example............................................................................................... 142
Troubleshooting .NET Applications................................................................................................. 143
Resolving the Error...............................................................................................................143
6
Contents
Pivotal GemFire Native Client
Using GemStone.GemFire.Cache.dll As a Private Assembly.............................................. 143
Implementing the Shared Assembly.....................................................................................144
Part VIII: Preserving Data....................................................................... 146
High Availability for Client-Server Communication..........................................................................147
Configuring Native Clients for High Availability.................................................................... 147
Sending Periodic Acknowledgment...................................................................................... 148
Enabling Queue Conflation to Improve Update Performance.........................................................148
Durable Client Messaging............................................................................................................... 149
Durable Client Messaging Requirements............................................................................. 149
Client-Side Configuration...................................................................................................... 150
Configuring a Durable Native Client.......................................................................... 150
Configuring Durable Interest in Keys.........................................................................151
Sending Cache Ready Messages to the Server.................................................................. 151
Disconnecting from the Server............................................................................................. 152
Life Cycle of a Durable Client.............................................................................................. 152
Initial Operation........................................................................................................ 152
Disconnection............................................................................................................. 152
Reconnection............................................................................................................. 153
Durable Message Replay.......................................................................................... 154
Application Operations During Interest Registration..................................................155
Implementing Cache Listeners for Durable Clients..............................................................155
Part IX: Security.......................................................................................158
Authentication.................................................................................................................................. 159
Process and Multiuser Authentication.................................................................................. 159
Configuring Credentials for Authentication..........................................................................160
Configuring Authentication by the Cache Server................................................................. 162
Server Authentication Errors................................................................................................ 162
Creating Multiple Secure User Connections........................................................................ 162
Requirements and Caveats for RegionService......................................................... 163
Using an LDAP Server for Client Authentication..................................................................163
Encrypted Authentication.................................................................................................................164
Encrypt Credentials with Diffe-Hellman................................................................................164
Using PKCS for Encrypted Authentication........................................................................... 165
Client Authorization......................................................................................................................... 166
Configuring Client Authorization........................................................................................... 166
Post-Operative Authorization................................................................................................ 166
Determining Pre- or Post-Operation Authorization..............................................................166
Security-Related System Properties (gfcpp.properties).................................................................. 167
SSL Client/Server Communication..................................................................................................168
Step 1: Download and install OpenSSL...............................................................................168
Step 2: Create keystores......................................................................................................168
Step 3: Configure environment variables............................................................................. 169
Step 4: Configure SSL properties in gfcpp.properties and gemfire.properties..................... 169
Step 5: Start and stop the client and server........................................................................ 169
Limitations............................................................................................................................. 170
Part X: Remote Querying........................................................................172
Remote Querying Basics.................................................................................................................173
Examples Data and Class Definitions.................................................................................. 173
Executing a Query from the Native Client........................................................................... 174
Querying the Portfolios Region............................................................................................ 175
7
Contents
Pivotal GemFire Native Client
Modifying Cache Contents................................................................................................... 177
Creating Indexes...................................................................................................................177
Remote Querying Requirements.......................................................................................... 178
Using Query Strings in the Native Client........................................................................................ 179
FROM Clause....................................................................................................................... 179
Using Iterator Variables............................................................................................. 180
Importing and Using Object Classes......................................................................... 181
Predefined Class Types.............................................................................................181
Specifying the Object Types of FROM Clause Collections....................................... 182
SELECT Projection List........................................................................................................ 183
SELECT Statement Query Results...................................................................................... 183
WHERE Clause.................................................................................................................... 185
Joins...................................................................................................................................... 185
Accessing Cached Data.................................................................................................................. 186
Basic Region Access............................................................................................................186
Attribute Visibility.................................................................................................................. 187
Modifying Query Scope........................................................................................................ 187
Nested Query Scopes.......................................................................................................... 188
When Names Cannot Be Resolved..................................................................................... 189
Query Language Elements..............................................................................................................189
Method Invocation................................................................................................................ 189
Query Language Literals Supported by Native Client.......................................................... 189
Type Conversions................................................................................................................. 190
Remote Query API.......................................................................................................................... 191
Creating and Managing Queries.......................................................................................... 191
Query Result Sets................................................................................................................ 192
Query Code Samples Returning ResultSet..........................................................................192
Query Code Samples Returning StructSet.......................................................................... 193
Part XI: Continuous Querying................................................................ 198
How Continuous Querying Works................................................................................................... 199
Implementing a Continuous Query..................................................................................................201
Configuring Your System for Continuous Querying............................................................. 202
Writing the Continuous Query.............................................................................................. 202
Writing the CQ Listener or CQ Status Listener....................................................................203
CqEvent Object.....................................................................................................................207
Running the Continuous Query Code.................................................................................. 208
CQ Execution Options.......................................................................................................... 209
When an Error Occurs in a Running CQ............................................................................. 210
Managing Continuous Queries........................................................................................................210
CQ API and Main Features............................................................................................................ 211
Part XII: Using Connection Pools.......................................................... 214
How Client Load Balancing Works................................................................................................. 215
Server Locators.................................................................................................................... 215
Connection Pools..................................................................................................................215
Discovering Locators Dynamically........................................................................................216
Configuring Pools............................................................................................................................ 217
Native Client Pool API..........................................................................................................217
Pool Configuration Example and Settings............................................................................217
Subscription Properties.........................................................................................................220
Running the Connection Pool Code.....................................................................................221
8
Contents
Pivotal GemFire Native Client
Part XIII: Transactions.............................................................................224
How Native Client Transactions Work............................................................................................ 225
Running a Native Client Transaction.............................................................................................. 226
Suspending and Resuming Transactions....................................................................................... 227
Part XIV: Function Execution................................................................. 230
Understanding Data-Aware Function Routing.................................................................................231
How Functions Execute...................................................................................................................231
Executing Functions in GemFire..................................................................................................... 236
Running the Function........................................................................................................... 236
Programming to Get Function Results................................................................................. 238
Solutions and Use Cases.....................................................................................................239
Part XV: Delta Propagation.....................................................................240
How Delta Propagation Works........................................................................................................241
Delta Propagation API.....................................................................................................................242
Cloning............................................................................................................................................. 242
Implementing Delta Propagation..................................................................................................... 243
Exceptions and Limitations..............................................................................................................244
Examples of Delta Propagation...................................................................................................... 246
Part XVI: Appendix A: Interoperability of Language Classes and
Types..................................................................................................... 254
C++ Class to .NET Class Mappings............................................................................................... 255
Interoperability of C++ Types When Using PDX Serialization............................................. 257
Java to .NET Type Mapping Table................................................................................................. 258
Part XVII: Appendix B: System Statistics............................................. 262
Sampling Statistics.......................................................................................................................... 263
System Performance Statistics....................................................................................................... 263
Region Statistics................................................................................................................... 263
Cache Performance Statistics.............................................................................................. 264
Continuous Query Statistics................................................................................................. 265
CQ Service Statistics............................................................................................................265
Pool Statistics....................................................................................................................... 265
Delta Statistics...................................................................................................................... 266
Operating System Statistics............................................................................................................ 267
Linux Process Statistics........................................................................................................267
Solaris Process Statistics..................................................................................................... 267
Windows Process Statistics..................................................................................................268
Part XVIII: Appendix C: Installing the SQLite Persistence Manager... 272
Linux Installation..............................................................................................................................273
Solaris Installation........................................................................................................................... 273
Windows Installation........................................................................................................................ 274
Part XIX: Glossary................................................................................... 276
9
Contacting Pivotal
Pivotal GemFire Native Client
Contacting Pivotal
Current Pivotal Customers
Submit a ticket from the Help & Support Page.
x
Contacting Pivotal
Pivotal GemFire Native Client
xi
Pivotal GemFire Native Client User's Guide
Pivotal GemFire Native Client
Part
I
Pivotal GemFire Native Client User's Guide
Revised January 30, 2015.
Pivotal GemFire Native Client User's Guide provides step-by-step
procedures for installation, configuration, and development of the
Pivotal® GemFire™ native client.
Intended Audience
Pivotal GemFire Native Client User's Guide is intended for anyone
who wants to use the Pivotal GemFire native client, including C++
and .NET developers programming their applications to use GemFire.
This guide assumes experience developing with C++ and .NET.
For information on how to manage and configure Pivotal GemFire
deployments, see the Pivotal GemFire User's Guide.
Document Roadmap
Pivotal GemFire Native Client User's Guide provides detailed
documentation on how to program and configure C++ and .NET client
applications that connect to GemFire servers. For information on how
to install and configure GemFire servers, see Pivotal GemFire User's
Guide.
Pivotal GemFire Native Client User's Guide contains the following
sections:
12
Pivotal GemFire Native Client User's Guide
Pivotal GemFire Native Client
13
Supported Configurations and System Requirements
Pivotal GemFire Native Client
Part
II
Supported Configurations and System
Requirements
Topics:
•
•
•
•
Pivotal GemFire Supported
Configurations
Pivotal GemFire Tools
Supported Configurations
Pivotal GemFire Native Client
Supported Configurations
Pivotal GemFire Modules
Supported Configurations
The sections that follow document supported operating system
platforms and describe additional system requirements for Pivotal
GemFire.
Revised January 27, 2015.
14
Pivotal GemFire Supported Configurations
Pivotal GemFire Native Client
Pivotal GemFire Supported Configurations
Pivotal GemFire® is supported on a variety of platforms.
The supported configurations tables are organized by the 32-bit and 64-bit platforms on which Pivotal
GemFire standalone runs with Level A support.
•
•
•
•
32-Bit Platforms on page 15
64-Bit Platforms on page 16
Production Support and Developer Support on page 16
Pure Java Mode and Other Configurations on page 16
Running Pivotal GemFire clusters with a mix of different platforms has not been tested. We recommend
that you use a consistent platform on all machines in your cluster.
Note: The Pivotal GemFire product download does not include Java. You must download and
install a supported JDK on each system running GemFire. GemFire requires the JDK (and not just
a JRE) to run gfsh commands and to launch servers and locators using the ServerLauncher or
LocatorLauncher APIs.
32-Bit Platforms
Operating System
Processor
Oracle Java
SE 6, Update
26 and Later
Oracle Java
SE 7, Update
5 and Later
Production or
Development
RHEL 5
x86
Yes
Yes
Production
RHEL 6
x86
Yes
Yes
Production
Solaris 9
SPARC
Yes
Yes
Production
Solaris 10
SPARC
Yes
Yes
Production
SLES 10**
x86
Yes
Yes
Production
Windows 2008
Server R1**
x86
Yes
Yes
Production
Windows 2008
Server R2
x86
Yes
Yes
Production
Windows 7
Enterprise
x86
Yes
Yes
Production
Windows 7
(versions other
than Enterprise)**
x86
Yes
Yes
Development
*The Microsoft Loopback Adapter is not supported.
**Indicates operating systems that have not been fully tested but are still supported.
15
Supported Configurations and System Requirements
Pivotal GemFire Native Client
64-Bit Platforms
Operating System
Processor
Oracle Java
SE 6, Update
26 and Later
Oracle Java
SE 7, Update
5 and Later
Production or Development
Mac OS X 10.8,
10.9**
x64
Yes
Yes
Development
RHEL 5
x64
Yes
Yes
Production
RHEL 6
x64
Yes
Yes
Production
Solaris 10
SPARC
Yes
Yes
Production
SLES 10**
x64
Yes
Yes
Production
Windows 2008
Server R1**
x64
Yes
Yes
Production
Windows 2008
Server R2
x64
Yes
Yes
Production
Windows 7
Enterprise
x64
Yes
Yes
Production
Windows 7
(versions other
than Enterprise)**
x64
Yes
Yes
Development
Ubuntu 10.x**
x64
Yes
Yes
Development
*The Microsoft Loopback Adapter is not supported.
**Indicates operating systems that have not been fully tested but are still supported.
Production Support and Developer Support
The tables indicate whether the supported configuration is for production or development. Generally,
production support means you can run your production application on the platform; developer support
means you can develop on the platform but you should not run your production application on it.
Many developers use consumer-focused operating systems that are not certified for production use,
and most Pivotal products function well on popular consumer operating systems. The developer support
designation is intended to convey which products are "known to work" and that Pivotal will provide besteffort support for resolving reported issues that are discovered. Developer Support certifications are not
supported for use in production.
Pure Java Mode and Other Configurations
If you want to run GemFire on a platform other than those listed in the two tables, contact your sales
representative. Pivotal will evaluate whether it can support the platform, either as-is or under a special
agreement.
For example, GemFire can also be installed to run in pure Java mode on any standard Java platform with
some functional differences. However, this type of configuration is not tested and should not be used for
production purposes without consulting with Pivotal support first. See Running Pivotal GemFire in Pure
Java Mode on page 18 for more information.
16
Supported Configurations and System Requirements
Pivotal GemFire Native Client
Host Machine Requirements
This topic describes the host machine requirements for Pivotal GemFire.
Each machine that will run Pivotal GemFire must meet the following requirements:
•
•
•
•
A supported Java SE installation. See Pivotal GemFire Supported Configurations on page 15.
Adequate per-user quota of file handles (ulimit for Solaris and Linux).
TCP/IP.
System clock set to the correct time and a time synchronization service such as Network Time Protocol
(NTP). Correct time stamps allow you to do the following activities:
•
•
Produce logs that are useful for troubleshooting. Synchronized time stamps ensure that log
messages from different hosts can be merged to reproduce a chronological history of a distributed
run.
• Aggregate product-level and application-level time statistics.
• Accurately monitor your GemFire system with scripts and other tools that read the system statistics
and log files.
Hostname and host files that are properly configured for the machine. Hostname and host file
configuration can affect gfsh and Pulse functionality. On Linux and Unix hosts, see the system
manpages for more information on setting hostname and hosts.
Note: For troubleshooting, you must run a time synchronization service on all hosts. Synchronized
time stamps allow you to merge log messages from different hosts, for an accurate chronological
history of a distributed run.
Dependencies on Linux RPM Packages
Pivotal GemFire has RPM package dependencies for the Red Hat Enterprise Linux (RHEL) 5 and 6
distributions.
The i386 or i686 after the package names indicates that you must install the package for that particular
architecture, regardless of the native operating system architecture. The packages listed are available on
the default media for each distribution.
Linux Version
glibc
libgcc
RHEL Server release 5 (i686)
glibc
libgcc
RHEL Server release 5 (x86_64)
glibc (i686)
libgcc (i386)
RHEL Server release 6 (i686)
glibc
libgcc
RHEL Server release 6 (x86_64)
glibc (i686)
libgcc (i386)
For versions of Linux not listed in the table, you can verify that you meet the Pivotal GemFire
dependencies at the library level with this command:
prompt> ldd <path to GemFire product dir>/lib/libgemfire.so
These libraries are external dependencies of the native library (libgemfire.so or libgemfire64.so). Check
that the output of ldd includes all of these:
libdl.so.2
libm.so.6
libpthread.so.0
17
Pivotal GemFire Tools Supported Configurations
Pivotal GemFire Native Client
libc.so.6
libgcc_s.so.1
For details on the ldd tool, see the online Linux man page for ldd.
Running Pivotal GemFire in Pure Java Mode
Pivotal GemFire can run in pure Java mode, meaning GemFire runs without the GemFire native code.
Running in pure Java mode enables Pivotal GemFire to run with some functional differences on
unsupported platforms. Unsupported platforms include any platforms that are not listed in Pivotal GemFire
Supported Configurations on page 15. Running in pure Java mode should not be used for production
environments without consulting with Pivotal support first. In pure Java mode, distributed system members
still have access to GemFire's caching and distribution capabilities, but the following features may be
disabled:
•
•
Operating system statistics. Platform-specific machine and process statistics such as CPU usage and
memory size.
Access to the process ID. Only affects log messages about the application. The process ID is set to
"0" (zero) in pure Java mode.
Pivotal GemFire Tools Supported Configurations
Pivotal GemFire Tools, including Pulse, GFMon, DataBrowser and VSD, are supported on a variety of
platforms.
Pivotal GemFire Tools, including Pulse, GFMon, DataBrowser and VSD, are supported on a variety of
platforms.
GemFire Pulse System Requirements
Verify that your system meets the installation and runtime requirements for GemFire Pulse.
The GemFire Pulse Web application has been tested for compatibility with the following Web browsers.
GemFire Pulse may work with other operating systems and browsers upon which it has not been tested:
•
•
•
•
Internet Explorer 9.0.8112.16421
Safari 5.1.7 for Windows
Google Chrome 22.0.1229.79 m
Mozilla Firefox 16.0.1
GemFire Pulse has been tested for standalone deployment on Tomcat and Jetty.
If you are interested in using a platform that is not listed, please contact your Pivotal sales representative.
Pivotal will evaluate whether it can support the platform, either as-is or under a special agreement, or
whether there are reasons that prevent deploying on a particular platform.
DataBrowser System Requirements
Verify that your system meets the installation and runtime requirements for DataBrowser.
DataBrowser is designed to run on Microsoft Windows and Red Hat Linux. DataBrowser requires the Java
Runtime Environment (JRE) 1.6.0 or higher to run.
18
Supported Configurations and System Requirements
Pivotal GemFire Native Client
You must use the Pivotal GemFire DataBrowser executable included with GemFire 7.0 to connect to
Pivotal GemFire 7.0 distributed systems. You cannot use past versions of Pivotal GemFire DataBrowser to
connect to a GemFire 7.0 system.
The minimum recommended hardware is a machine with Intel® 2 GHz Pentium® 4 (or equivalent) and 2
GB RAM.
In the DataBrowser startup scripts, you may want to set the JVM heap size explicitly to a value suitable for
the typical result sizes returned by the queries that you run in your distributed system.
VSD System Requirements
View a list of platforms that are known to work with VSD.
The following platforms are known to work with the current VSD release:
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
RedHat 3
RedHat 4
RedHat 5
RedHat 6
SLES 8
SLES 9
SLES 10
Solaris 2.8
Mac OSX
Windows NT 4.0
Windows 98/ME/2K
Windows XP SP3
Windows Server 2003
Windows Server 2008
Windows 7 (Note: Requires slight modification; see below.)
64-Bit Platform Support
VSD is a 32-bit application. If you are running VSD on a 64-bit operating system, you may need to install
32-bit OS libraries to run the application if they are not already installed. On Linux, to find out which
libraries are missing you can try running the following:
ldd <product_dir>/tools/vsd/bin/vsdwishLinux
For 64-bit Windows, you can modify the scripts and executables as described in the note below.
Windows 7 and Later Support
To use VSD on Windows 7, perform the following steps:
1. Start Windows Explorer and navigate to the GemFireProductDir\tools\vsd\bin\ directory
(where GemFireProductDir corresponds to the location where you installed GemFire.)
2. Right click and select properties for vsd.bat.
3. Select the Compatibility tab.
4. Click "Run this program in compatibility mode for" and then select Windows XP SP3.
5. Repeat steps 2 and 3 for all the other executables in the bin directory.
19
Pivotal GemFire Native Client Supported Configurations
Pivotal GemFire Native Client
Pivotal GemFire Native Client Supported
Configurations
The Pivotal GemFire native client provides access for C++ and Microsoft® .NET™ clients to the GemFire
distributed system. It operates on platforms running Microsoft Windows, Linux (Intel), and Sun Solaris.
Operating system requirements are listed in the chart below:
Operating System RAM
Swap Space
32-bit Disk Space
Required
64-Bit Disk Space
Required
Solaris 9
2GB
256MB
73MB
81MB
Solaris 10
2GB
256MB
73MB
81MB
RHEL 5
2GB
256MB
100MB
114MB
RHEL 6
2GB
256MB
100MB
114MB
SLES 10**
2GB
256MB
100MB
114MB
Windows 2008
Server R1**
2GB
256MB
102MB
114MB
Windows 2008
Server R2
2GB
256MB
102MB
114MB
Windows 7
Enterprise
2GB
256MB
102MB
114MB
**Indicates operating systems that have not been fully tested but are still supported.
Windows Support Details
The Pivotal GemFire native client is built and tested on Windows 2008 R2.
The native client is not supported on Windows NT 4.0.
Runtime Library Requirements
The GemFire native client also requires the Microsoft Visual C++ 2010 Redistributable Package. This
package is installed for you when you use the native client .msi installer. If you do not use the .msi installer,
you will need to install this package for your platform architecture manually (x86 or x64) on all machines
that will run the GemFire native client. This package contains runtime libraries needed by the native client.
.NET Framework Version Support
The Pivotal GemFire native client is supported with Microsoft .NET Framework 2.0, 3.0, and 3.5 and 4.0.
A Microsoft .NET Framework must be installed to support the C++/CLI (Common Language Infrastructure)
library for the native client.
Note: You can download the .NET Framework Redistributable Package (x86 for 32-bit or x64 for
64-bit) for the version you wish to use from http://www.microsoft.com/downloads. If it isn’t listed on
the Download Center page, use the Search tool to search for ".NET Framework".
The Pivotal GemFire native client 7.0 now supports .NET 4.0 and Visual Studio 2010 (for compiling C
++ applications on Windows). For advantages and more information on the features of .NET 4.0 and
Visual Studio 2010 SP1, see http://msdn.microsoft.com/en-us/library/dd831853(v=vs.100).aspx and http://
msdn.microsoft.com/en-us/library/vstudio/w0x726c2(v=vs.100).aspx.
20
Supported Configurations and System Requirements
Pivotal GemFire Native Client
Linux
The Pivotal GemFire native client is built on Red Hat Enterprise ES 3, kernel version 2.4.21-47.EL.
The native client is tested on the following Linux versions:
•
•
Red Hat Enterprise 5 release 5 (Tikanga), kernel version 2.6.18-8.EL5
Red Hat Enterprise 6, kernel version 2.6.32-220.el6.x86_64
If you are not sure of the kernel version on your system, use this command to list it:
prompt> uname -r
The following table lists the RPM package dependencies for several Linux distributions. The i386 or
i686 after the package name indicates that you must install the package for that particular architecture
regardless of the native operating system architecture. All of the packages listed are available with the
default media for each distribution.
Table 1: GemFire Dependencies on Linux RPM Packages
Linux Version
glibc
libgcc
Red Hat Enterprise Linux
Server release 5 (i686)
glibc
libgcc
Red Hat Enterprise Linux
Server release 5 (x86_64)
glibc (i686)
libgcc (i386)
Red Hat Enterprise Linux
Server release 6 (i686)
glibc
glibc
Red Hat Enterprise Linux
Server release 6 (x86_64)
glibc (i686)
ibgcc (i386)
For versions of Linux not listed in the table, you can verify that you meet the native client dependencies at
the library level by using the ldd tool and entering this command:
prompt> ldd $GFCPP/lib/libgfcppcache.so
This step assumes you have already installed the native client and have set the GFCPP environment
variable to productDir, where productDir represents the location of the NativeClient_xxxx_bnnnnn directory
(xxxx is the four-digit product version and bnnnnn is the product build number).
The following libraries are external dependencies of the native library, libgfcppcache.so. Verify that the
ldd tool output includes all of these:
•
•
•
•
•
libdl.so.2
libm.so.6
libpthread.so.0
libc.so.6
libz.so.1
For details on the ldd tool, see its Linux online man page.
Solaris
The Pivotal GemFire native client is supported on the following Solaris versions:
•
•
Solaris 9 kernel update 118558-38
Solaris 10 kernel update 118833-24
21
Pivotal GemFire Modules Supported Configurations
Pivotal GemFire Native Client
Running Pivotal GemFire Native Client on vSphere
Without tuning, the Pivotal GemFire native client can suffer a performance drop in virtual environments,
including the VMware vSphere virtual platform. Without correct tuning, you can expect to see significant
performance degradation when running the GemFire native client on vSphere versus running GemFire on
dedicated hardware.
We recommend that you tune your GemFire native client on vSphere deployments using the same
guidelines published for Pivotal GemFire.
See "Improving GemFire Performance on vSphere" in the Pivotal GemFire User's Guide for a list of
guidelines and recommendations.
Pivotal GemFire native client has been certified and tested on vSphere 5.0.
Software Requirements for Using SSL
If you plan on using SSL in your GemFire native client and server deployment, you will need to download
and install OpenSSL.
The GemFire native client requires OpenSSL 1.0.1h or later. For Windows platforms, you can use either
the regular or OpenSSL "Light" version.
In addition, make sure that your system environment variables have been configured to include OpenSSL.
See "SSL Client/Server Communications" in the Pivotal GemFire Native Client User's Guide for
instructions.
Pivotal GemFire Modules Supported Configurations
This section describes supported configurations for the HTTP Session Management and Hibernate
modules.
HTTP Session Management Module Supported Configurations
Pivotal GemFire HTTP Session Management modules support the following application servers and their
versions.
Supported Application
Server
Supported
Version
Pivotal GemFire Module Documentation
tc Server
2.5
HTTP Session Management Module for Pivotal tc
Server
tc Server
2.6
HTTP Session Management Module for Pivotal tc
Server
tc Server
2.7
HTTP Session Management Module for Pivotal tc
Server
tc Server
2.8
HTTP Session Management Module for Pivotal tc
Server
tc Server
2.9
HTTP Session Management Module for Pivotal tc
Server
Tomcat
6.0
HTTP Session Management Module for Tomcat
Tomcat
7.0
HTTP Session Management Module for Tomcat
WebLogic
11g (10.3.x)
HTTP Session Management Module for AppServers
22
Supported Configurations and System Requirements
Pivotal GemFire Native Client
Supported Application
Server
Supported
Version
Pivotal GemFire Module Documentation
WebSphere
7, 8
HTTP Session Management Module for AppServers
JBoss
5, 6, 7
HTTP Session Management Module for AppServers
Note: This table lists all application server versions that have been tested with the GemFire HTTP
Session Management modules. However, the generic HTTP Session Management Module for
AppServers is implemented as a servlet filter and should work on any application server platform
that supports the Java Servlet 2.4 specification.
For more information on how to install and get started with the HTTP Session Management Modules, see
HTTP Session Management Quick Start.
Hibernate Module Supported Configurations
Pivotal GemFire supports Hibernate 3.3 and later (up to version 3.6.10) for use with the Hibernate Cache
module.
Note: GemFire does not currently support Hibernate 4.x for use with the Hibernate Cache module.
For more information on using the Hibernate Module, see "Hibernate Cache Module" in the Pivotal
GemFire User's Guide.
23
Getting Started with Pivotal GemFire Native Client
Pivotal GemFire Native Client
Part
III
Getting Started with Pivotal GemFire Native Client
Topics:
•
•
•
•
About the Native Client
Installing the Native Client
Running Native Client
Applications
QuickStart Examples and Guide
Getting Started with Pivotal GemFire Native Client gives you a
conceptual overview of the Pivotal™ GemFire® native client. It shows
you how to install the product, build native client applications on
various platforms, and run the product examples.
The Pivotal GemFire native client provides access for C++ and
Microsoft® .NET™ clients to the Pivotal GemFire distributed system.
24
About the Native Client
Pivotal GemFire Native Client
About the Native Client
The native client delivers the full set of capabilities supplied by Java clients communicating with the
GemFire cache server.
The native client is written entirely in C++, so its initialization process does not involve the creation
of a Java virtual machine. The .NET native client provides native operations for the .NET Framework
application developer who writes in .NET languages and needs to access the GemFire cache server.
Pivotal GemFire native clients in C++, Java, and .NET languages communicate only with the cache
server and do not communicate with each other. The native clients interface with the server at the sockets
level and implement the same wire protocol to the server. These capabilities produce extremely high
performance and system scalability.
C++ and .NET native clients provide access to the full region API, including support for application plugins, managed connectivity, highly available data, and reliable failover to a specified server list. All of this is
transparent to the end user.
You can configure GemFire native clients to cache data locally, or they can act in a cacheless mode where
they retrieve data from a cache server and directly pass it to other system members without incurring the
caching overhead. They can be configured as read only caches, or be configured to receive notifications
from the server whenever a key of interest to the client changes on the server.
This is a conceptual overview of how .NET and C++ applications access the cache server.
Figure 1: Pivotal GemFire Native Client Overview
Installing the Native Client
Procedures for installing the native client vary according to your operating system.
25
Getting Started with Pivotal GemFire Native Client
Pivotal GemFire Native Client
Installing on Linux and Solaris
You install the native client on Linux and Solaris by extracting the contents of a ZIP file. After installation,
you add environment settings.
Installation Prerequisites
Before you install the Pivotal GemFire native client, you must complete the following prerequisites:
•
•
•
Confirm that your system meets the hardware and software requirements described in Pivotal GemFire
Native Client Supported Configurations on page 20.
From the Pivotal GemFire product page, select Downloads.
On the Pivotal GemFire product download page, select and download the Pivotal GemFire native
client .zip file appropriate for your operating system and hardware architecture.
•
•
For Linux platforms, download
Pivotal_GemFire_NativeClient_Linux_architecture_xxxx_bNNNN.zip where
architecture equals either 32bit or 64bit, xxxx represents the four-digit version of the product and
bNNNN corresponds to the build number for the release.
For Solaris platforms, download
Pivotal_GemFire_NativeClient_Solaris_architecture_xxxx_bNNNN.zip where
architecture equals either 32bit or 64bit, xxxx represents the four-digit version of the product and
bNNNN corresponds to the build number for the release.
Installation Path for Linux
The default installation path is NativeClient_Linux_architecture_xxxx_bNNNN and reflects the
architecture, product version and build number identified in the filename of the installer .zip file.
Installation Path for Solaris
The default installation path is NativeClient_Solaris_architecture_xxxx_bNNNN and reflects the
architecture, product version and build number identified in the filename of the installer .zip file.
Environment Settings
When the installation is complete, you add these environment settings:
•
•
•
Set the GFCPP environment variable to productDir, where productDir represents the
NativeClient__platform_architecture_xxxx directory (where platform is the OS you are
installing upon, architecture is 32bit or 64bit, and xxxx is the four-digit product version).
Add $GFCPP/bin to the PATH.
Add $GFCPP/lib to the LD_LIBRARY_PATH.
Installing on Windows
Install the native client by using the Windows installer GUI interface or the command-line.
This installation topic contains the following sections:
•
•
•
•
Prerequisites on page 27
Install with the Windows Installer on page 27
Install from the MSI Command-Line on page 27
Repair a Windows Installation on page 28
26
Getting Started with Pivotal GemFire Native Client
Pivotal GemFire Native Client
Prerequisites
Before you install the Pivotal GemFire native client, you must complete the following prerequisites:
•
•
•
Confirm that your system meets the hardware and software requirements described in Windows
Support Details on page 20.
From the Pivotal GemFire product page, select Downloads.
On the Pivotal GemFire product download page, select and download the Pivotal GemFire native
client .msi file appropriate for your hardware architecture.
Note: If you do not install the native client using the .msi installer or the msiexec command-line,
then you will need to manually perform the configuration steps (such as configure your GFCPP and
PATH environment variables) and install the Microsoft Visual C++ 2010 Redistributable Package
described in the next section.
Install with the Windows Installer
The native client can be installed on Windows by using the
Pivotal_GemFire_NativeClient_architecture_xxxx_bNNNN.msi Windows installer, where
architecture is 32bit or 64bit, xxxx represents the four-digit version of the product and bNNNN corresponds
to the build number for the release. The installer requires msiexec version 3.0 or higher.
Note: You must be logged in with administrative privileges or the MSI installation will fail.
Double-click the MSI file to start the installation. You are prompted to enter the installation location of the
native client.
To complete the installation, the MSI installer automatically configures these native client system
environment settings:
•
•
•
Sets the GFCPP environment variable to productDir, where productDir represents the location where
you installed the native client.
Adds %GFCPP%\bin to the Windows PATH.
Installs Microsoft Visual C++ 2010 Redistributable Package for your platform architecture (x86 or x64).
This package contains runtime libraries required by the native client.
Install from the MSI Command-Line
In addition to the standard Windows installer interface, the
Pivotal_GemFire_NativeClient_architecture_xxxx_bNNNN.msi file provides a set of
command-line installer options. This is an example of the installer command-line syntax:
msiexec /q /i Pivotal_GemFire_NativeClient_architecture_xxxx_bNNNN.msi
DEFAULT_INSTALLDIR=<path>
The following table lists common msiexec command-line options for use with
NativeClient_architecture_xxxx.msi, along with an explanation of their usage. For a list of all
command-line options, enter msiexec/?.
Table 2: MSI Command-Line Options
Option
Explanation
/q
Creates a quiet installation with no interface or
prompts.
/i
Indicates that the product is to be installed or
configured.
27
Getting Started with Pivotal GemFire Native Client
Pivotal GemFire Native Client
Option
Explanation
DEFAULT_INSTALLDIR=<path>
Specifies the destination directory, if different from
the default.
/x
Indicates a product uninstall procedure.
Repair a Windows Installation
If problems occur with your Windows native client installation you can initiate a repair operation to restore
any missing elements or registry settings.
1. Double-click the MSI file, then click Next in the Setup Wizard screen.
2. In the following screen, click the Repair button.
3. In the next screen, click the Repair button.
Uninstalling the Native Client
How you uninstall the native client depends on your operating system.
Uninstall on Linux and Solaris
On Linux and Solaris, delete the NativeClient_Linux_architecture_xxxx_bNNNN product
directory or the NativeClient_Solaris_architecture_xxxx_bNNNN product directory and all of its
subdirectories.
Uninstall on Windows
Use the MSI installer graphical interface, its command-line interface, or the Windows Control Panel.
Uninstall from MSI Installer Graphical Interface
1. Double-click the MSI file, then click Next in the Setup Wizard screen.
2. In the following screen, click the Remove button.
3. In the next screen, click the Remove button and then click Finish.
Uninstall from MSI Installer Command-Line
This is the MSI command-line syntax for a quiet uninstall:
msiexec /q /x Pivotal_GemFire_NativeClient_architecture_xxxx_bNNNN.msi
DEFAULT_INSTALLDIR=<path>
Table 2: MSI Command-Line Options on page 27 lists common msiexec command-line options for
use with Pivotal_GemFire_NativeClient_architecture_xxxx_bNNNN.msi, along with an
explanation of their usage. For a list of all command-line options, enter msiexec/?.
The /x option uninstalls the product.
Uninstall from the Control Panel
1. From the Control Panel, open Add or Remove Programs.
2. Select Pivotal GemFire Native Client [32bit or 64bit] [x.x.x.x] (with the values appropriate for your
installation.)
3. Click the Remove button.
28
Running Native Client Applications
Pivotal GemFire Native Client
Running Native Client Applications
Set up the environment for the native client on multiple platforms. Compile and run client programs.
Developing C++ Programs on Linux
This section describes how to build and run a native client application on Linux.
Note: When compiling external projects or applications that are used or referenced by the GemFire
native client, make sure that you compile them for the same target architecture as your native client
installation. For example, if you installed the 32-bit (x86) version of the native client, compile your
external projects for 32-bit (x86) architecture.
Step 1: Set Environment Variables
Set the native client environment variables on each Linux host. For each case, productDir is the path to the
native client product directory.
For Bourne and Korn shells (sh, ksh, bash)
GFCPP=<productDir>;export GFCPP
PATH=$GFCPP/bin:$PATH;export PATH
LD_LIBRARY_PATH=$GFCPP/lib:$LD_LIBRARY_PATH;export LD_LIBRARY_PATH
Step 2: Compile C++ Clients and Dynamically Link Them to the GemFire Library
On Linux, the g++ compiler is supported. To build and link a C++ client to GemFire on Linux, the
compilation command line must include the arguments listed in the following table.
Table 3: Compiler Arguments (Linux)
Argument
Explanation
-D_REENTRANT
Required to compile Linux programs in a threadsafe way.
-m32
Enables 32-bit compilation.
-m64
Enables 64-bit compilation.
-I$GFCPP/include
Specifies the native client include directory.
The following table lists the linker switches that must be present on the command line when dynamically
linking to the GemFire library.
Table 4: Linker Switches (Dynamically Linking on Linux)
Argument
Explanation
-rpath $GFCPP/lib
Tells the linker to look in $GFCPP/lib for libraries
on which the GemFire library depends.
-L$GFCPP/lib
Tells the linker where to find the named libraries.
-o durableclient
Tells the linker to output an object file named
'durableclient'.
29
Getting Started with Pivotal GemFire Native Client
Pivotal GemFire Native Client
Argument
Explanation
-lgfcppcache
Links the GemFire C++ cache library to the
compiled executable.
The following examples compile and link the $GFCPP/SampleCode/quickstart/cpp/
DurableClient.cpp client to the durableclient output file.
Compiling and Dynamically Linking on Linux for 32-bit
g++ \
-D_REENTRANT \
-03 \
-Wall \
-m32 \
-I$GFCPP/include \
cpp/DurableClient.cpp \
cpp/plugins/DurableCacheListener.cpp \
-o cpp/DurableClient \
-L$GFCPP/lib \
-Wl,-rpath,$GFCPP/lib \
-lgfcppcache
Compiling and Dynamically Linking on Linux for 64-bit
g++ \
-D_REENTRANT \
-03 \
-Wall \
-m64 \
-I$GFCPP/include \
cpp/DurableClient.cpp \
cpp/plugins/DurableCacheListener.cpp \
-o cpp/DurableClient \
-L$GFCPP/lib \
-Wl,-rpath,$GFCPP/lib \
-lgfcppcache
Step 3: Make Sure the GemFire Library Can Be Loaded
When the C++ application is dynamically linked to the GemFire library, the library must be dynamically
loadable. To verify that the GemFire library is available for loading, make sure you have added the path
GemFireDir/lib to the LD_LIBRARY_PATH environment variable, where GemFireDir is the path to the
GemFire product directory.
Developing C++ Programs on Solaris
This section describes how to build and run a native client application on Solaris.
Note: When compiling external projects or applications that are used or referenced by the GemFire
native client, make sure that you compile them for the same target architecture as your native client
installation. For example, if you installed the 32-bit (x86) version of the native client, compile your
external projects for 32-bit (x86) architecture.
Step 1. Set Environment Variables
Set the native client environment variables on each Solaris host. For each case, productDir is the path to
the native client product directory.
30
Getting Started with Pivotal GemFire Native Client
Pivotal GemFire Native Client
For Bourne and Korn shells (sh, ksh, bash)
GFCPP= productDir ; export GFCPP
PATH=$GFCPP/bin:$PATH;export PATH
LD_LIBRARY_PATH=$GFCPP/lib:$LD_LIBRARY_PATH;export LD_LIBRARY_PATH
Step 2. Compile C++ Clients and Dynamically Link to Them to GemFire Library
Version 5.9 of the SUNpro compiler is supported on Solaris. The linker switches vary according to whether
you are statically linking to the GemFire library.
To build and link a C++ client to GemFire on Solaris, the compilation command line must include the
appropriate arguments from this table.
Table 5: Compiler Arguments on Solaris
Argument
Explanation
-D_REENTRANT
Required to compile Solaris programs in a threadsafe way.
-xarch=v8plus
Enables 32-bit compilation.
-xarch=v9
Enables 64-bit compilation.
-ldl; -lpthread; -lc; -lm; -lsocket; -lrt; lnsl; -ldemangle; -lkstat; -lz
Additional libraries.
-library=stlport4
Solaris library compilation.
-I$ GFCPP /include
Specifies the GemFire include directory.
Step 3. Make Sure the GemFire Library Can Be Loaded
When a C++ application is not statically linked to the GemFire library, the library must be dynamically
loadable.
To verify that the GemFire library is available for loading, make sure you have added the path GemFireDir/
lib to the LD_LIBRARY_PATH environment variable, where GemFireDir is the path to the GemFire
product directory.
Developing C++ Programs on Windows
GemFire uses the Visual Studio 2010 Service Pack 1 compiler for C++ programs on Windows, which
invokes Microsoft® cl.exe from the command line at compile time.
The Pivotal GemFire native client 7.0 now supports .NET 4.0 and Visual Studio 2010. For advantages and
more information on the features of .NET 4.0 and Visual Studio 2010 SP1, see http://msdn.microsoft.com/
en-us/library/dd831853(v=vs.100).aspx and http://msdn.microsoft.com/en-us/library/vstudio/
w0x726c2(v=vs.100).aspx.
Visual Studio 2010 SP1 is the recommended compiler. If you are using any other compiler, contact
technical support for assistance.
Note: When compiling external projects or applications that are used or referenced by the GemFire
native client, make sure that you compile them for the same target architecture as your native client
installation. For example, if you installed the 32-bit (x86) version of the native client, compile your
external projects for 32-bit (x86) architecture.
31
Getting Started with Pivotal GemFire Native Client
Pivotal GemFire Native Client
Step 1. Installer Sets Up Environment Variables
When you install on Windows, the installer performs these tasks:
•
•
Sets the GFCPP environment variable to productDir, where productDir is the path to the native client
product directory.
Adds the %GFCPP%\bin executable directory to the Windows PATH.
Step 2. Choose 32-bit or 64-bit Command-line Prompt
For 32-bit:
Start > Programs > Microsoft Visual Studio 2010 > Visual Studio Tools > Visual Studio 2010
Command Prompt
For 64-bit:
Start > Programs > Microsoft Visual Studio 2010 > Visual Studio Tools > Visual Studio 2010 x64
Win64 Command Prompt
To build using the Microsoft Visual Studio Interface, from the Solutions Platform, choose Win32 or x86
from the Build menu for 32-bit builds or x64 for a 64-bit build.
Step 3. Compile C++ Clients and Dynamically Link Them to GemFire Library
The following table lists the compiler and linker switches that must be present on the cl.exe command
line.
Note: If you want to use the Visual Studio user interface instead of invoking cl.exe from the
command line, be sure to supply these parameters.
Table 6: Compiler and Linker Switches for Windows
Argument
Explanation
/MD
Memory model.
/EHsc
Catches C++ exceptions only and tells the compiler
to assume that *extern* C functions never throw a
C++ exception.
/GR
Runtime type information.
-I%GFCPP%\include
Specifies the GemFire include directory.
%GFCPP%\lib\gfcppcache.lib
Specifies the library file for the shared library.
/D_CRT_SECURE_NO_DEPRECATE
Suppresses warnings. Required for Visual Studio
2010.
/D_CRT_NON_CONFORMING_SWPRINTFS
Suppresses warnings. Required for Visual Studio
2010.
Step 4. Verify that You Can Load the GemFire Library
Because GemFire does not provide a library that can be linked statically into an application on Windows,
you must dynamically link to the GemFire library.
To make the GemFire library available for loading, verify that the directory GemFireDir/bin is included in
the PATH environment variable, where GemFireDir is the path to the GemFire product directory.
32
QuickStart Examples and Guide
Pivotal GemFire Native Client
QuickStart Examples and Guide
Run the GemFire native client QuickStart examples and read the QuickStart Guide to understand native
client functionality.
Get a conceptual overview of the VMware® Pivotal™ GemFire® native client. Install the product, build
native client applications on various platforms, and run the QuickStart examples.
Running the QuickStart Examples
The QuickStart examples demonstrate the capabilities of the GemFire native client, and provide source
code so you can examine how each example is designed. C++ and C# examples demonstrate how the
native client performs as a C++ or C# client.
Read the QuickStart Guide first (the README.HTML file included with the QuickStart examples) to learn
more about running the QuickStart examples.
QuickStart examples are packaged as part of the native client product download listed on the Pivotal
GemFire download page. Product examples are installed into the following directory when you install the
Pivotal GemFire native client:
<native_client_installdir>/SampleCode/quickstart
where native_client_installdir is the location where you installed the Pivotal GemFire native client.
Note: The C# examples have a .config file that should not be modified or deleted. The file
provides references to support files for running the example.
Accessing the QuickStart Guide
The QuickStart Guide for the native client consists of a set of compact programming samples that
demonstrate both C++ and C# client operations.
Open the QuickStart Guide to rapidly become familiar with native client functionality. To view the
QuickStart Guide, navigate to the SampleCode/quickstart directory of your Pivotal GemFire native
client installation, and open the README.HTML file. The README.html file describes each programming
sample, explains initial environment setup, and provides instructions for running the QuickStart examples.
33
Setting System Properties
Pivotal GemFire Native Client
Part
IV
Setting System Properties
Topics:
•
•
•
Setting System Properties describes how to configure Pivotal GemFire
native clients and cache servers to participate in a distributed system.
Configuring the Native Client
and Cache Server
Attributes in gfcpp.properties
gfcpp.properties Example File
34
Configuring the Native Client and Cache Server
Pivotal GemFire Native Client
Configuring the Native Client and Cache Server
You can configure clients through files and API calls. The servers are configured through command-line
input and configuration files.
See Pivotal GemFire User's Guide for information on installing and configuring GemFire cache servers.
Native Client Configuration
You configure the native client in two files: gfcpp.properties for native client system-level configuration
and cache.xml for cache-level configuration.
The configuration of the caches is part of the application development process. See Cache Initialization
File. (The cache-level configuration file is generally referred to as cache.xml, but you can use any name.)
•
•
•
•
About gfcpp.properties Configuration File on page 35
Configuration File Locations on page 35
Configuring System Properties for the Native Client on page 35
Running a Native Client Out of the Box on page 36
About gfcpp.properties Configuration File
The gfcpp.properties file provides local settings required to connect a client to a distributed system,
along with settings for licensing, logging, and statistics. See Attributes in gfcpp.properties.
The application software may include a set of gfcpp.properties files. You set any attributes needed for
the application design in these files, then you can add any attributes needed for the local site.
If you do not have gfcpp.properties files, use any text editor to create them. See gfcpp.properties
Example File for a sample of the file format and contents.
Configuration File Locations
A native client looks for gfcpp.properties first in the working directory where the process runs, then
in productDir/defaultSystem. Use the defaultSystem directory to group configuration files or to
share them among processes for more convenient administration. If gfcpp.properties is not found, the
process starts up with the default settings.
For the cache.xml cache configuration file, a native client looks for the path specified by the cachexml-file attribute in gfcpp.properties (see Attributes in gfcpp.properties). If the cache.xml is not
found, the process starts with an unconfigured cache.
Configuring System Properties for the Native Client
The typical configuration procedure for a native client includes the high-level steps listed below. The rest of
this chapter provides the details.
1. Place the gfcpp.properties file for the application in the working directory or in productDir/
defaultSystem. Use the configuration file that came with the application software if there is one, or
create your own. See gfcpp.properties Example File for a sample of the file format and contents.
2. Place the cache.xml file for the application in the desired location and specify its path in the
gfcpp.properties file.
3. Add other attributes to the gfcpp.properties file as needed for the local system architecture. See
Attributes in gfcpp.properties for the configurable attributes, and gfcpp.properties Example File for a
sample of the file format.
35
Setting System Properties
Pivotal GemFire Native Client
Running a Native Client Out of the Box
If you start a native client without any configurations, it uses any attributes set programmatically plus any
hard-coded defaults (listed in Attributes in gfcpp.properties). Running with the defaults is a convenient way
to learn the operation of the distributed system and to test which attributes need to be reconfigured for your
environment.
Cache Server Configuration
You configure the cache server in two files: gemfire.properties for server system-level configuration
and cache.xml for cache-level configuration.
The configuration of the caches is part of the application development process. See Cache Initialization
File. (The cache-level configuration file is generally referred to as cache.xml, but you can use any name.)
Configuration File Locations
For the GemFire cache server, the gemfire.properties file is usually stored in the current working
directory. For more information, see the GemFire User's Guide.
For the cache.xml cache configuration file, a native client looks for the path specified by the cachexml-file attribute in gfcpp.properties (see Attributes in gfcpp.properties). If the cache.xml is not
found, the process starts with an unconfigured cache.
Modifying Attributes Outside the gemfire.properties File
In addition to the gemfire.properties file, you can pass attributes to the cache server on the
command line. These override any settings found in the gemfire.properties file for starting the cache
server.
For example, the attributes are passed in as attName=attValue pairs, such as logfile=newLogFileName. For more information, see the GemFire User's Guide.
Attribute Definition Priority
You can specify attributes in different ways, which can cause conflicting definitions. Applications can be
configured programmatically, and that has priority over other settings.
Check your application documentation to see whether this applies in your case.
•
•
•
•
•
Programmatic configuration
Properties set at the command line
currentWorkingDirectory/gfcpp.properties file
productDir/defaultSystem/gfcpp.properties file
GemFire defaults
In case an attribute is defined in more than one place, the first source in this list is used:
The gfcpp.properties files and programmatic configuration are optional. If they are not present, no
warnings or errors occur. For details on programmatic configuration through the Properties object, see
Defining Properties Programmatically.
For information on the GemFire cache server configuration, see the GemFire User's Guide.
36
Attributes in gfcpp.properties
Pivotal GemFire Native Client
Search Path for Multiple gfcpp.properties Files
The native client and cache server processes first look for their properties file in the productDir/
defaultSystem directory, then in the working directory.
Any properties set in the working directory override settings in the defaultSystem/gfcpp.properties
file.
If you are running multiple processes on one machine, you can configure the gfcpp.properties file
in the defaultSystem directory as a shared file that all processes can find. If a few processes need a
slightly different configuration, you can put individual gfcpp.properties files in their home directories to
override specific properties.
Overriding gfcpp.properties Settings
Application developers have the option of configuring system attributes programmatically, rather than using
the gfcpp.properties file.
Attributes set programmatically override any matching attribute settings in the gfcpp.properties
file, but additional attributes not set programmatically will be configured using the settings in
gfcpp.properties.
Defining Properties Programmatically
You can pass in specific gfcpp properties programmatically by using a Properties object to define the
non-default properties.
Example:
PropertiesPtr systemProps = Properties::create();
systemProps->insert( "statistic-archive-file", "stats.gfs" );
systemProps->insert( "cache-xml-file", "./myapp-cache.xml" );
systemProps->insert( "stacktrace-enabled", "true" );
CacheFactoryPtr systemPtr = CacheFactory::createCacheFactory(systemProps);
Attributes in gfcpp.properties
A variety of gfcpp.properties settings can be used when a native client connects to a distributed
system.
The following settings can be configured:
•
•
•
•
•
General Properties Basic information for the process, such as cache creation parameters.
Logging Properties How and where to log system messages.
Statistics Archiving Properties How to collect and archive statistics information.
Durable Client Properties Information about the durable clients connected to the system.
Security Properties Information about various security parameters.
37
Setting System Properties
Pivotal GemFire Native Client
Attribute Definitions
The following tables list GemFire configuration attributes that can be stored in the gfcpp.properties file
to be read by a native client.
For the system properties that relate to high availability, see Sending Periodic Acknowledgement. For a
list of security-related system properties and their descriptions, see the table System Properties for Client
Authentication and Authorization.
Table 7: Attributes in gfcpp.properties: General Properties
gfcpp.properties Attribute
Description
Default
appdomain-enabled
If true , allows native client
to work when multiple .NET
appdomains are in use.
false
cache-xml-file
Name and path of the file whose no default
contents are used by default to
initialize a cache if one is created.
If not specified, the native client
starts with an empty cache, which
is populated at runtime.
See Cache Initialization File for
more information on the cache
initialization file.
heap-lru-delta
When heap LRU is triggered, this 10
is the amount that gets added to
the percentage that is above the
heap-lru-limit amount. LRU
continues until the memory usage
is below heap-lru-limit
minus this percentage. This
property is only used if heaplru-limit is greater than 0 .
heap-lru-limit
Maximum amount of memory,
in megabytes, used by the
cache for all regions. If this limit
is exceeded by heap-lrudelta percent, LRU reduces the
memory footprint as necessary.
If not specified, or set to 0 ,
memory usage is governed by
each region's LRU entries limit, if
any.
0
conflate-events
Client side conflation setting,
which is sent to the server.
server
connect-timeout
Amount of time (in seconds) to
wait for a response after a socket
connection attempt.
59
connection-pool-size
Number of connections per
endpoint
5
crash-dump-enabled
Whether crash dump generation
for unhandled fatal errors is
true
38
Setting System Properties
gfcpp.properties Attribute
Pivotal GemFire Native Client
Description
Default
enabled. True is enabled, false
otherwise.
disable-chunk-handler-thread
When set to false, each
application thread processes its
own response. If set to true, the
chunk-handler-thread processes
the response for each application
thread.
false
disable-shuffling-of-endpoints
If true, prevents server endpoints
that are configured in pools from
being shuffled before use.
false
grid-client
If true, the client does not start
various internal threads, so that
startup and shutdown time is
reduced.
false
max-fe-threads
Thread pool size for parallel
2 * number of CPU cores
function execution. An example of
this is the GetAll operations.
max-socket-buffer-size
Maximum size of the socket
buffers, in bytes, that the native
client will try to set for clientserver connections.
notify-ack-interval
Interval, in seconds, in which
1
client sends acknowledgments for
subscription notifications.
notify-dupcheck-life
Amount of time, in seconds,
the client tracks subscription
notifications before dropping the
duplicates.
ping-interval
Interval, in seconds, between
10
communication attempts with
the server to show the client is
alive. Pings are only sent when
the ping-interval elapses
between normal client messages.
This must be set lower than
the server's maximum-timebetween-pings.
redundancy-monitor-interval
Interval, in seconds, at which the
subscription HA maintenance
thread checks for the configured
redundancy of subscription
servers.
10
stacktrace-enabled
If true, the exception classes
capture a stack trace that
can be printed with their
printStackTrace function.
If false , the function prints
false
65 * 1024
300
39
Setting System Properties
gfcpp.properties Attribute
Pivotal GemFire Native Client
Description
Default
a message that the trace is
unavailable.
tombstone-timeout
Time in milliseconds used to
timeout tombstone entries when
region consistency checking is
enabled.
480000
Note: Do not modify this
property unless instructed
by technical support.
Table 8: Attributes in gfcpp.properties: Logging Properties
gfcpp.properties Attribute
Description
Default
log-disk-space-limit
Maximum amount of disk space,
in megabytes, allowed for all log
files, current, and rolled. If set to
0 , the space is unlimited.
0
log-file
Name and full path of the file
where a running client writes
log messages. If not specified,
logging goes to stdout.
no default file
log-file-size-limit
Maximum size, in megabytes, of
a single log file. Once this limit
is exceeded, a new log file is
created and the current log file
becomes inactive. If set to 0 , the
file size is unlimited.
0
log-level
Controls the types of messages
that are written to the
application's log. These are
the levels, in descending order
of severity and the types of
message they provide:
config
•
•
•
•
•
Error (highest severity) is
a serious failure that will
probably prevent program
execution.
Warning is a potential
problem in the system.
Info is an informational
message of interest to
the end user and system
administrator.
Config is a static configuration
message, often used to debug
problems with particular
configurations.
Fine, Finer, Finest, and
Debug provide tracing
information. Only use these
40
Setting System Properties
gfcpp.properties Attribute
Pivotal GemFire Native Client
Description
Default
with guidance from technical
support.
Enabling logging at any level
enables logging for all higher
levels.
Table 9: Attributes in gfcpp.properties: Statistics Archiving Properties
gfcpp.properties Attribute
Description
Default
statistic-sampling-enabled
Controls whether the process
creates a statistic archive file.
true
statistic-archive-file
Name and full path of the file
where a running system member
writes archives statistics. If
archive-disk-spacelimit is not set, the native
client appends the process ID
to the configured file name, like
statArchive-PID.gfs. If the
space limit is set, the process
ID is not appended but each
rolled file name is renamed to
statArchive-ID.gfs , where ID is
the rolled number of the file.
./statArchive.gfs
archive-disk-space-limit
Maximum amount of disk space, 0
in megabytes, allowed for all
archive files, current, and rolled. If
set to 0 , the space is unlimited.
archive-file-size-limit
Maximum size, in bytes, of a
0
single statistic archive file. Once
this limit is exceeded, a new
statistic archive file is created and
the current archive file becomes
inactive. If set to 0 , the file size is
unlimited.
statistic-sample-rate
Rate, in seconds, that statistics
are sampled. Operating system
statistics are updated only when
a sample is taken. If statistic
archival is enabled, then these
samples are written to the
archive.
1
Lowering the sample rate
for statistics reduces system
resource use while still providing
some statistics for system tuning
and failure analysis.
You can view archived statistics
with the optional VSD utility.
41
gfcpp.properties Example File
Pivotal GemFire Native Client
gfcpp.properties Attribute
Description
Default
enable-time-statistics
Enables time-based statistics
false
for the distributed system and
caching. For performance
reasons, time-based statistics are
disabled by default. See System
Statistics.
Table 10: Attributes in gfcpp.properties: Durable Client Properties
gfcpp.properties Attribute
Description
Default
auto-ready-for-events
Whether client subscriptions
automatically receive events
when declaratively configured
via XML. If set to false ,
event startup is not automatic
and you need to call the
Cache.ReadyForEvents()
method API after subscriptions
for the server to start delivering
events.
true
durable-client-id
Identifier to specify if you want the empty
client to be durable.
durable-timeout
Time , in seconds, a durable
300
client's subscription is maintained
when it is not connected to the
server before being dropped.
Table 11: Attributes in gfcpp.properties: Security Properties
gfcpp.properties Attribute
Description
Default
security-client-dhalgo
Diffie-Hellman secret key
algorithm.
null
security-client-kspath
keystore (.pem file ) path.
null
security-client-auth-factory
Factory method for the security
AuthInitialize module.
empty
security-client-auth-library
Path to the client security library
for the AuthInitialize
module.
empty
ssl-keystore-password
Keystore password.
null
gfcpp.properties Example File
Use the gfcpp.properties file to configure distributed system connections for the Pivotal GemFire native
client.
The following example shows the format of a gfcpp.properties file. The first two attributes in this example
should be set by programmers during application development, while other attributes are set on-site during
42
Setting System Properties
Pivotal GemFire Native Client
system integration. The properties and their default settings that can be set in this file are described in
detail in Attributes in gfcpp.properties on page 37.
gfcpp.properties File Format
#Tue Feb 14 17:24:02 PDT 2006
log-level=info
cache-xml-file=./cache.xml
stacktrace-enabled=true
Using the Default Sample File
A sample gfcpp.properties file is included with the Pivotal GemFire native client installation in the
productDir/defaultSystem directory.
To use this file:
1. Copy the file to the directory where you start the application.
2. Uncomment the lines you need and edit the settings as shown in this example:
cache-xml-file=test.xml
3. Start the application.
Default gfcpp.properties File
# Default Gemfire C++ distributed system properties
# Copy to current directory and uncomment to override defaults.
#
## Debugging support, enables stacktraces in gemfire::Exception.
#
# The default is false, uncomment to enable stacktraces in
exceptions.
#stacktrace-enabled=true
#crash-dump-enabled=true
#
#
## Cache region configurtion
#
#cache-xml-file=cache.xml
#
## Log file config
#
#log-file=gemfire_cpp.log
#log-level=config
# zero indicates use no limit.
#log-file-size-limit=0
# zero indicates use no limit.
#log-disk-space-limit=0
#
## Statistics values
#
# the rate is in seconds.
#statistic-sample-rate=1
#statistic-sampling-enabled=true
#statistic-archive-file=statArchive.gfs
# zero indicates use no limit.
#archive-file-size-limit=0
# zero indicates use no limit.
43
Setting System Properties
Pivotal GemFire Native Client
#archive-disk-space-limit=0
#enable-time-statistics=false
#
## Heap based eviction configuration
#
# maximum amount of memory used by the cache for all regions, 0
disables this feature
#heap-lru-limit=0
# percentage over heap-lru-limit when LRU will be called.
#heap-lru-delta=10
#
## Durable client support
#
#durable-client-id=
#durable-timeout=300
#
## SSL socket support
#
#ssl-enabled=false
#ssl-keystore=
#ssl-truststore=
#
## .NET AppDomain support
#
#appdomain-enabled=false
#
## Misc
#
#conflate-events=server
#disable-shuffling-of-endpoints=false
#grid-client=false
#max-fe-threads=
#max-socket-buffer-size=66560
# the units are in seconds.
#connect-timeout=59
#notify-ack-interval=10
#notify-dupcheck-life=300
#ping-interval=10
#redundancy-monitor-interval=10
#auto-ready-for-events=true
#
## module name of the initializer pointing to sample
## implementation from templates/security
#security-client-auth-library=securityImpl
## static method name of the library mentioned above
#security-client-auth-factory=createUserPasswordAuthInitInstance
## credential for Dummy Authenticator configured in server.
## note: security-password property will be inserted by the
initializer
## mentioned in the above property.
#security-username=root
44
Setting System Properties
Pivotal GemFire Native Client
45
Configuring the Native Client Cache
Pivotal GemFire Native Client
Part
V
Configuring the Native Client Cache
Topics:
•
•
•
•
•
•
•
Caches
Cache Initialization File
(cache.xml)
Regions
Region Entries
Region Consistency
Region Attributes
Cache Management
Configuring the Native Client Cache describes the Pivotal GemFire
native client cache functionality, initialization file, and APIs. It provides
information about creating and working with caches, cache regions,
and region entries.
The Pivotal GemFire native client cache provides a framework for
native clients to store, manage, and distribute application data.
46
Caches
Pivotal GemFire Native Client
Caches
The cache is the entry point to native client data caching in GemFire. Through the cache, native clients
gain access to the GemFire caching framework for data loading, distribution, and maintenance.
About the Native Client Cache
The cache consists of data regions, each of which can contain any number of entries. Region entries hold
the cached data. Every entry has a key that uniquely identifies it within the region and a value where the
data object is stored.
The Cache instance allows your process to set general parameters for communication between a cache
and other caches in the distributed system, and to create and access any region in the cache.
Regions are created from the Cache instance. Regions provide the entry point to the interfaces for
instances of Region and RegionEntry.
Main Features and Functionality
The native client cache provides the following features and functionality.
•
•
•
•
•
•
Local and distributed data caching for fast access.
Data distribution between applications on the same or different platforms.
Local and remote data loading through application plug-ins.
Application plug-ins for synchronous and asynchronous handling of data events.
Automated and application-specific data eviction for freeing up space in the cache, including optional
overflow to disk.
System message logging, and statistics gathering and archiving.
For more information specific to your client programming language, see the online API documentation and
Pivotal GemFire Native Client C++ API and Pivotal GemFire Native Client .NET Caching API.
Cache APIs
The native client has two cache APIs, RegionService and Cache.
RegionService API
RegionService provides:
•
•
Access to existing cache regions.
Access to the standard query service for the cache, which sends queries to the servers. See Remote
Querying and Continuous Querying.
RegionService is inherited by Cache.
You do not use instances of RegionService except for secure client applications with many users. See
Creating Multiple Secure User Connections with RegionService.
Cache API
Use the Cache to manage your client caches. You have one Cache per client.
The Cache inherits RegionService and adds management of these client caching features:
47
Configuring the Native Client Cache
•
•
•
•
Pivotal GemFire Native Client
Region creation.
Subscription keepalive management for durable clients.
Access to the underlying distributed system.
RegionService creation for secure access by multiple users.
Local, Remote, and Distributed Caches
The distributed system defines how native client and cache server processes find each other.
The distributed system keeps track of its membership list and makes its members aware of the identities of
the other members in the distributed system.
A cache for a native client is referred to as its local cache. All other caches in the distributed system are
considered remote caches to the application. Every cache server and application process has its own
cache. The term distributed cache is used to describe the union of all caches in a GemFire distributed
system.
Creating and Accessing a Cache
When you create a native client cache, you are creating a native client cache instance. You must provide
some basic configuration information such as a connection name and cache initialization parameters for
the native client's cache instance.
When you create a cache, you provide the following input:
•
•
Connection name. Used in logging to identify both the distributed system connection and the cache
instance. If you do not specify a connection name, a unique (but non-descriptive) default name is
assigned.
cache.xml to initialize the cache (if the initialization is not done programmatically). To
modify the cache structure, you just edit cache.xml in your preferred text editor. No changes to the
application code are required. If you do not specify a cache initialization file, you need to initialize the
cache programmatically.
The cache.xml file contains XML declarations for cache, region, and region entry configuration.
This XML declares server connection pools and regions:
<cache>
<region name="clientRegion1" refid="PROXY">
<region-attributes pool-name="serverPool1"/>
</region>
<region name="clientRegion2" refid="PROXY">
<region-attributes pool-name="serverPool2"/>
</region>
<region name="localRegion3" refid="LOCAL"/>
<pool name="serverPool1">
<locator host="host1" port="40404"/>
</pool>
<pool name="serverPool2">
<locator host="host2" port="40404"/>
</pool>
</cache>
When you use the regions, the client regions connect to the servers through the pools named in their
configurations.
48
Cache Initialization File (cache.xml)
Pivotal GemFire Native Client
This file can have any name, but is generally referred to as cache.xml .
For a list of the parameters in the cache.xml file, including the DTD, see Cache Initialization File.
To create your cache, call the CacheFactory create function. The cache object it returns gives access
to the native client caching API. For example:
CacheFactoryPtr cacheFactory = CacheFactory::createCacheFactory();
CachePtr cachePtr = cacheFactory->create();
Note: For more information on how to create a cache, see Pivotal GemFire Native Client C++ API
on page 86, Pivotal GemFire Native Client C# .NET API on page 112, and the native client API
reference (https://www.vmware.com/support/pubs/vfabric-apiref.html).
Closing the Cache
Use the Cache::close function to release system resources when you finish using the cache.
After the cache is closed, any further method calls on the cache or any region object results in a
CacheClosedException.
If the cache is in a durable client, you need to use the keepalive version of the close method. See
Disconnecting From the Server.
Cache Initialization File (cache.xml)
To ease the task of managing the structure of the cache, you can define the default Pivotal GemFire cache
structure in an XML-based initialization file.
To modify the cache structure, you just edit cache.xml in your preferred text editor. No changes to the
application code are required.
This chapter describes the file format of the cache.xml file and discusses its contents.
Cache Initialization File Basics
The contents of the cache initialization file are used to populate or update a cache.
This occurs when a cache server starts up, when a client application explicitly creates its cache, or when a
client explicitly loads a new structure into an existing cache.
The initialization file can have any name, but is generally referred to as cache.xml. Both client
applications and cache servers can use an optional cache.xml file to ease the initialization process.
File Contents
The contents of a declarative XML file correspond to APIs declared in the Cache.hpp and Region.hpp
header files. The cache initialization file allows you to accomplish declaratively many of the cache
management activities that you can program through the API.
•
•
The contents of the cache initialization file must conform to the XML definition in productDir/dtd/
gfcpp-cache7000.dtd (see Native Client Cache XML DTD).
The name of the declarative XML file is specified when establishing a connection to the distributed
system. You can define it by setting the cache-xml-file configuration attribute in the
49
Configuring the Native Client Cache
Pivotal GemFire Native Client
gfcpp.properties file for the native client. For details about the gfcpp.properties file, see
Setting System and Cache Properties.
Example cache.xml File
An example cache.xml file shows cache and region initialization for a client. The example shows only a
subset of the possible data configurations.
For detailed information about cache and region configuration, including the default attribute settings, see
The Native Client Cache. Also see the online API documentation for Cache and RegionAttributes.
For information a cache with server pool, see Using Connection Pools. The example below shows a
cache.xml file that creates two regions.
•
•
Region region1 is defined with a full set of region attributes and application plug-ins. The region's
entries have RegionTimeToLive and RegionIdleTimeout expiration attributes set (Expiration
Attributes).
Region region2 uses mostly default values.
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE cache PUBLIC
"-//GemStone Systems, Inc.//GemFire Declarative Caching 7.0//EN"
"http://www.gemstone.com/dtd/gfcpp-cache7000.dtd">
<!-- Sample cache.xml file -->
<!-- Example Declarative Cache Initialization with cache.xml -->
<cache>
<pool name="examplePool" subscription-enabled="true">
<server host="localhost" port="24680" />
</pool>
<region name="root1" refid="CACHING_PROXY">
<region-attributes pool-name="examplePool"
initial-capacity="25"
load-factor="0.32"
concurrency-level="10"
lru-entries-limit="35">
<region-idle-time>
<expiration-attributes timeout="20" action="destroy"/>
</region-idle-time>
<entry-idle-time>
<expiration-attributes timeout="10" action="invalidate"/>
</entry-idle-time>
<region-time-to-live>
<expiration-attributes timeout="5" action="local-destroy"/>
</region-time-to-live>
<entry-time-to-live>
<expiration-attributes timeout="10" action="localinvalidate"/>
</entry-time-to-live>
</region-attributes>
</region>
</cache>
For details about the individual region attributes, see Region Attributes.
50
Configuring the Native Client Cache
Pivotal GemFire Native Client
Native Client Cache XML DTD
The contents of the cache initialization file must conform to the data type definitions in gfcppcache7000.dtd.
The gfcpp-cache7000.dtd file is in the productDir/dtd directory of your native client installation.
The following example presents the file gfcpp-cache7000.dtd, which defines the XML used by the
GemFire native client for declarative caching. The DTD file identifies the valid element tags that may
be present in your XML file, the attributes that correspond to each element, and the valid values for the
elements and attributes.
<?xml version="1.0" encoding="UTF-8"?>
<!-This is the XML DTD for the GemFire Native Client declarative caching XML
file.
The contents of a declarative XML file correspond to APIs found in the
Gemfire Native Client product, more specifically in the
Cache.hpp and Region.hpp files in the product include directory
A declarative caching XML file is used to populate a Cache
when it is created.
-->
<!-The "client-cache" element is the root element of the declarative cache
file.
This element configures a GemFire Native Client Cache and describes the
root regions it contains, if any.
-->
<!ELEMENT client-cache (pool*, root-region*, region*, pdx?)>
<!ATTLIST client-cache
endpoints CDATA #IMPLIED
redundancy-level CDATA #IMPLIED
>
<!-A "locator" element specifies the host and port that a server locator is
listening on
-->
<!ELEMENT locator EMPTY>
<!ATTLIST locator
host CDATA #REQUIRED
port CDATA #REQUIRED
>
<!-A "server" element specifies the host and port that a cache server is
listening on
-->
<!ELEMENT server EMPTY>
<!ATTLIST server
host CDATA #REQUIRED
port CDATA #REQUIRED
51
Configuring the Native Client Cache
Pivotal GemFire Native Client
>
<!-- A "pdx" element specifies the properties related to pdx types -->
<!ELEMENT pdx EMPTY>
<!ATTLIST
ignore-unread-fields (false | true) #IMPLIED
read-serialized (false | true) #IMPLIED
>
<!-A "pool" element specifies a client-server connection pool.
-->
<!ELEMENT pool (locator+|server+)>
<!ATTLIST pool
free-connection-timeout
CDATA #IMPLIED
load-conditioning-interval
CDATA #IMPLIED
min-connections
CDATA #IMPLIED
max-connections
CDATA #IMPLIED
retry-attempts
CDATA #IMPLIED
idle-timeout
CDATA #IMPLIED
ping-interval
CDATA #IMPLIED
name
CDATA #REQUIRED
read-timeout
CDATA #IMPLIED
server-group
CDATA #IMPLIED
socket-buffer-size
CDATA #IMPLIED
subscription-enabled
(false | true) #IMPLIED
subscription-message-tracking-timeout CDATA #IMPLIED
subscription-ack-interval
CDATA #IMPLIED
subscription-redundancy
CDATA #IMPLIED
statistic-interval
CDATA #IMPLIED
pr-single-hop-enabled
(true | false) #IMPLIED
thread-local-connections
(false | true) #IMPLIED
multiuser-authentication
(false | true) #IMPLIED
>
<!-A root-region" element describes a root region whose entries and
subregions will be stored in memory.
Note that the "name" attribute specifies the simple name of the region;
it cannot contain a "/".
-->
<!ELEMENT root-region (region-attributes?, region*)>
<!ATTLIST root-region
name CDATA #REQUIRED
>
<!-A "region" element describes a region (and its entries) in GemFire
Native Client cache. Note that the "name" attribute specifies the simple
name of the region; it cannot contain a "/".
-->
<!ELEMENT region (region-attributes?, region*)>
<!ATTLIST region
name CDATA #REQUIRED
refid CDATA #IMPLIED
>
<!-A "region-attributes" element describes the attributes of a region to
52
Configuring the Native Client Cache
Pivotal GemFire Native Client
be created. For more details see the RegionFactory header in the
product include directory
-->
<!ELEMENT region-attributes ((region-time-to-live |
region-idle-time | entry-time-to-live | entry-idle-time |
partition-resolver |
cache-loader | cache-listener | cache-writer | persistence-manager)*)>
<!ATTLIST region-attributes
caching-enabled (true | TRUE | false | FALSE) #IMPLIED
cloning-enabled (true | TRUE | false | FALSE) #IMPLIED
scope (local | distributed-no-ack | distributed-ack ) #IMPLIED
initial-capacity CDATA #IMPLIED
load-factor CDATA #IMPLIED
concurrency-level CDATA #IMPLIED
lru-entries-limit CDATA #IMPLIED
disk-policy (none | overflows | persist ) #IMPLIED
endpoints CDATA #IMPLIED
client-notification (true | TRUE | false | FALSE) #IMPLIED
pool-name CDATA #IMPLIED
concurrency-checks-enabled (true | TRUE | false | FALSE) #IMPLIED
id CDATA #IMPLIED
refid CDATA #IMPLIED
>
<!-A "region-time-to-live" element specifies a Region's time to live
-->
<!ELEMENT region-time-to-live (expiration-attributes)>
<!-A "region-idle-time" element specifies a Region's idle time
-->
<!ELEMENT region-idle-time (expiration-attributes)>
<!-A "entry-time-to-live" element specifies a Region's entries' time to
live
-->
<!ELEMENT entry-time-to-live (expiration-attributes)>
<!-A "entry-idle-time" element specifies a Region's entries' idle time
-->
<!ELEMENT entry-idle-time (expiration-attributes)>
<!-A "properties" element specifies a persistence properties
53
Configuring the Native Client Cache
Pivotal GemFire Native Client
-->
<!ELEMENT properties (property*)>
<!-An "expiration-attributes" element describes expiration
-->
<!ELEMENT expiration-attributes EMPTY>
<!ATTLIST expiration-attributes
timeout CDATA #REQUIRED
action (invalidate | destroy | local-invalidate | local-destroy) #IMPLIED
>
<!-A "cache-loader" element describes a region's CacheLoader
-->
<!ELEMENT cache-loader EMPTY >
<!ATTLIST cache-loader
library-name CDATA #IMPLIED
library-function-name CDATA #REQUIRED
>
<!-A "cache-listener" element describes a region's CacheListener
-->
<!ELEMENT cache-listener EMPTY>
<!ATTLIST cache-listener
library-name CDATA #IMPLIED
library-function-name CDATA #REQUIRED
>
<!-A "cache-writer" element describes a region's CacheListener
-->
<!ELEMENT cache-writer EMPTY>
<!ATTLIST cache-writer
library-name CDATA #IMPLIED
library-function-name CDATA #REQUIRED
>
<!-A "partition-resolver" element describes a region's PartitionResolver
-->
<!ELEMENT partition-resolver EMPTY>
<!ATTLIST partition-resolver
library-name CDATA #IMPLIED
library-function-name CDATA #REQUIRED
>
<!-A "persistence-manager" element describes a region's persistence feature
-->
54
Regions
Pivotal GemFire Native Client
<!ELEMENT persistence-manager (properties)>
<!ATTLIST persistence-manager
library-name CDATA #IMPLIED
library-function-name CDATA #REQUIRED
>
<!-A "property" element describes a persistence property
-->
<!ELEMENT property EMPTY>
<!ATTLIST property
name CDATA #REQUIRED
value CDATA #REQUIRED
>
Regions
You create cache regions either programmatically or through declarative statements in the cache.xml
file. Generally, a cache is organized and populated through a combination of the two approaches.
A distributed region can be either non-partitioned or a partitioned region. See the GemFire User's Guide for
detailed descriptions of both non-partitioned and partitioned regions. Region creation is subject to attribute
consistency checks. The requirements for consistency between attributes are detailed both in the online
API documentation and throughout the discussion of Region Attributes.
See the GemFire User's Guide for a complete description of consistency checking in Pivotal GemFire.
Declarative Region Creation
Declarative region creation involves placing the region's XML declaration, with the appropriate attribute
settings, in the cache.xml file that is loaded at cache creation.
Note: Before creating a region, specify region attributes. See Region Attributes on page 64.
Regions are placed inside the cache declaration in region elements. For example:
<cache>
<pool name="examplePool" subscription-enabled="true" >
<server host="localhost" port="40404" />
</pool>
<region name="A" refid="PROXY">
<region-attributes pool-name="examplePool"/>
</region>
<region name="A1">
<region-attributes refid="PROXY" pool-name="examplePool"/>
</region>
<region name="A2" refid="CACHING_PROXY">
<region-attributes pool-name="examplePool">
<region-time-to-live>
<expiration-attributes timeout="120" action="invalidate"/>
</region-time-to-live>
</region-attributes>
</region>
55
Configuring the Native Client Cache
Pivotal GemFire Native Client
</cache>
The cache.xml file contents must conform to the DTD provided in the productDir/dtd/gfcppcache7000.dtd file. For details, see Cache Initialization File.
Note: For more information on how to create a region, see Pivotal GemFire Native Client C++ API
on page 86, Pivotal GemFire Native Client C# .NET API on page 112, and the native client API
reference (https://www.vmware.com/support/pubs/vfabric-apiref.html).
Programmatic Region Creation
You create regions programmatically with the regionFactory class.
Note: Before creating a region, specify region attributes. See Region Attributes on page 64.
Create your regions using the regionFactory class.
C++ RegionFactory Example
RegionFactoryPtr regionFactory =
cachePtr->createRegionFactory(CACHING_PROXY);
RegionPtr regPtr0 = regionFactory->setLruEntriesLimit(20000)
->create("exampleRegion0");
Note: For more information on how to create a region, see Pivotal GemFire Native Client C++ API
on page 86, Pivotal GemFire Native Client C# .NET API on page 112, and the native client API
reference (https://www.vmware.com/support/pubs/vfabric-apiref.html).
Invalidating and Destroying Regions
Invalidation marks all entries contained in the region as invalid (with null values). Destruction removes the
region and all of its contents from the cache.
You can execute these operations explicitly in the local cache in the following ways:
•
•
Through direct API calls from the native client.
Through expiration activities based on the region's statistics and attribute settings.
In either case, you can perform invalidation and destruction as a local or a distributed operation.
•
•
•
A local operation affects the region only in the local cache.
A distributed operation works first on the region in the local cache and then distributes the operation
to all other caches where the region is defined. This is the proper choice when the region is no longer
needed, or valid, for any application in the distributed system.
If the region on the server is configured as a partitioned region, it cannot be cleared using API calls from
the native client.
A user-defined cache writer can abort a region destroy operation. Cache writers are synchronous listeners
with the ability to abort operations. If a cache writer is defined for the region anywhere in the distributed
system, it is invoked before the region is explicitly destroyed.
Region invalidation and destruction can cause other user-defined application plug-ins to be invoked as
well. These plug-ins are described in detail in Application Plug-Ins.
56
Region Entries
Pivotal GemFire Native Client
Whether carried out explicitly or through expiration activities, invalidation and destruction cause event
notification.
Region Access
You can use Cache::getRegion to retrieve a reference to a specified region.
RegionPtr returns NULL if the region is not already present in the application's cache. A server region
must already exist.
A region name cannot contain these characters:
•
•
•
•
•
•
•
•
•
<
>
:
"
/
\
|
?
*
Getting the Region Size
The Region API provides a size method (Size property for .NET) that gets the size of a region.
For native client regions, this gives the number of entries in the local cache, not on the servers.
See the Region online API documentation for details
Region Entries
Region entries hold cached application data. Entries are automatically managed according to region
attribute settings.
You can create, update, invalidate, and destroy entries through explicit API calls or through operations
distributed from other caches.
When the number of entries is very large, a partitioned region can provide the required data management
capacity if the total size of the data is greater than the heap in any single JVM. See the GemFire User's
Guide for a detailed description of partitioned regions and related data management.
When an entry is created, a new object is instantiated in the region containing:
•
•
The entry key.
The entry value. This is the application data object. The entry value may be set to NULL , which is the
equivalent of an invalid value.
Entry operations invoke callbacks to user-defined application plug-ins. In this chapter, the calls that may
affect the entry operation itself (by providing a value or aborting the operation, for example) are highlighted,
but all possible interactions are not listed. For details, see Application Plug-Ins.
DateTime objects must be stored in the cache in UTC, so that times correspond between client and
server. If you use a date with a different time zone, convert it when storing into and retrieving from the
cache.
57
Configuring the Native Client Cache
Pivotal GemFire Native Client
Entry Distribution Requirements
Entry data distributed among members of the distributed system must be serializable. Entry keys and
values are serialized for distribution.
If a native client defines a region, it must register any serializable types for all classes of objects stored
in the region. This includes entries that the application gets or puts, as well as entries that are pushed to
the client's cache automatically through distribution. The types must be registered before the native client
connects to the distributed system.
See Serializing Data on page 91 for more information about these requirements.
Registering Interest for Entries
For native client regions, you can programmatically register interest in entry keys stored on a cache server
region. A client region receives update notifications from the cache server for the keys of interest.
You can register interest for specific entry keys or for all keys. Regular expressions can be used to register
interest for keys whose strings match the expression. You can also unregister interest for specific keys,
groups of keys based on regular expressions, or for all keys.
Note: Interest registration and unregistration are symmetrical operations. Consequently, you
cannot register interest in all keys and then unregister interest in a specific set of keys. Also, if you
first register interest in specific keys with registerKeys, then call registerAllKeys, you must
call unregisterAllKeys before specifying interest in specific keys again.
•
•
•
•
•
•
Client API for Registering Interest on page 58
Setting Up Client Notification on page 58
Registering Interest for Specific Keys on page 59
Registering Interest for All Keys on page 59
Registering Interest Using Regular Expressions on page 59
Register Interest Scenario on page 60
Client API for Registering Interest
You register client interest through the C++ or NET API. The C++ API provides the registerKeys,
registerAllKeys, and registerRegex methods, with corresponding unregistration accomplished
using the unregisterKeys, unregisterAllKeys, and unregisterRegex methods. The .NET
API provides the RegisterKeys, RegisterAllKeys, and RegisterRegex methods, with
corresponding unregistration accomplished using the UnregisterKeys, UnregisterAllKeys, and
UnregisterRegex methods.
The registerKeys, registerRegex and registerAllKeys methods have the option to populate
the cache with the registration results from the server. The registerRegex and registerAllKeys
methods can also optionally return the current list of keys registered on the server.
Setting Up Client Notification
In addition to the programmatic function calls, to register interest for a server region and receive updated
entries you need to configure the region with the PROXY or CACHING_PROXY RegionShortcut
setting. The region's pool should have subscription-enabled=true set either in the client XML or
programmatically via a CacheFactory::setSubscriptionEnabled(true) API call. Otherwise, when
you register interest, you will get an UnsupportedOperationException.
<region name = "listenerWriterLoader" refid="CACHING_PROXY">
58
Configuring the Native Client Cache
Pivotal GemFire Native Client
...
Both native clients and Java clients that have subscriptions enabled track and drop (ignore) any duplicate
notifications received. To reduce resource usage, a client expires tracked sources for which new
notifications have not been received for a configurable amount of time.
Notification Sequence
Notifications invoke CacheListeners of cacheless clients in all cases for keys that have been registered
on the server. Similarly, invalidates received from the server invoke CacheListeners of cacheless
clients.
If you register to receive notifications, listener callbacks are invoked irrespective of whether the key is in
the client cache when a destroy or invalidate event is received.
Registering Interest for Specific Keys
You register and unregister interest for specific keys through the registerKeys and unregisterKeys
functions. You register interest in a key or set of keys by specifying the key name using the programmatic
syntax shown in the following example:
keys0.push_back(keyPtr1);
keys1.push_back(keyPtr3);
regPtr0->registerKeys(keys0);
regPtr1->registerKeys(keys1);
The programmatic code snippet in the next example shows how to unregister interest in specific keys:
regPtr0->unregisterKeys(keys0);
regPtr1->unregisterKeys(keys1);
Registering Interest for All Keys
If the client registers interest in all keys, the server provides notifications for all updates to all keys in the
region. The next example shows how to register interest in all keys:
regPtr0->registerAllKeys();
regPtr1->registerAllKeys();
The following example shows a code sample for unregistering interest in all keys.
regPtr0->unregisterAllKeys();
regPtr1->unregisterAllKeys();
Registering Interest Using Regular Expressions
The registerRegex function registers interest in a regular expression pattern. The server automatically
sends the client changes for entries whose keys match the specified pattern.
Keys must be strings in order to register interest using regular expressions.
The following example shows interest registration for all keys whose first four characters are Key-,
followed by any string of characters. The characters .* represent a wildcard that matches any string.
regPtr1->registerRegex("Key-.*");
59
Configuring the Native Client Cache
Pivotal GemFire Native Client
To unregister interest using regular expressions, you use the unregisterRegex function. The next
example shows how to unregister interest in all keys whose first four characters are Key-, followed by any
string (represented by the .* wildcard).
regPtr1->unregisterRegex("Key-.*");
Register Interest Scenario
In this register interest scenario, a cache listener is used with a cacheless region that has
subscription-enabled set to true. The client region is configured with caching disabled; client
notification is enabled; and a cache listener is established. The client has not registered interest in any
keys:
When a value changes in another client, it sends the event to the server. The server will not send the event
to the cacheless client, even though client-notification is set to true.
To activate the cache listener so the cacheless region receives updates, the client should explicitly register
interest in some or all keys by using one of the API calls for registering interest. This way, the client
receives all events for the keys to which it has registered interest. This applies to Java-based clients as
well as native clients.
Using serverKeys to Retrieve a Set of Region Keys
You can retrieve the set of keys defined in the cache server process that are associated with the client
region by using the Region::serverKeys API function. If the server region is defined as a replicate, the
keys returned consist of the entire set of keys for the region.
The following example shows how the client can programmatically call serverKeys.
VectorOfCacheableKey keysVec;
region-<serverKeys( keysVec );
size_t vlen = keysVec.size();
bool foundKey1 = false;
bool foundKey2 = false;
for( size_t i = 0; i < vlen; i++ ) {
CacheableStringPtr strPtr = dynCast<CacheableStringPtr< keysVec.at( i );
std::string veckey = strPtr-<asChar();
if ( veckey == "skey1" ) {
printf( "found skey1" );
foundKey1 = true;
}
if ( veckey == "skey2" ) {
printf( "found skey2" );
foundKey2 = true;
}
}
An UnsupportedOperationException occurs if the client region is not a native client region. A
MessageException occurs if the message received from the server could not be handled, which can
occur if an unregistered typeId is received in the reply.
Adding Entries to the Cache
A region is populated with cached entries in several ways:
•
Explicitly, when an application executes a create or a put operation for a single entry or for multiple
entries that do not already exist in the cache.
60
Configuring the Native Client Cache
•
•
Pivotal GemFire Native Client
Implicitly, when a client does a get on a single entry or on multiple entries that do not already exist in
the cache. In this case, the entry is retrieved from a remote cache or through a cache loader (Under
Specifying Application Plug-In Attributes on page 74see "CacheLoader"). A client can also use
getAll to populate a region with all values for an array of keys. See Accessing Entries.
Automatically, when entries are created in remote caches.
If any cache writer is available in the distributed region, it is called before the entry is created and it can
abort the creation process.
Note: For more information on how to create entries, see Pivotal GemFire Native Client C++ API
on page 86, Pivotal GemFire Native Client C# .NET API on page 112, and the native client API
reference (https://www.vmware.com/support/pubs/vfabric-apiref.html).
Adding Entries to the Local Cache
If just the local cache is to be populated, you can either create an entry using the localCreate Region
API, or put an entry using localPut. See the C++ and .NET online API documentation for details about
localCreate and localPut.
DateTime objects must be stored in the cache in UTC, so that times correspond between client and
server. If you use a date with a different time zone, convert it when storing into and retrieving from the
cache. This example converts a local time to UTC for a put operation:
DateTime t1( 2009, 8, 13, 4, 11, 0, DateTimeKind.Local);
region0.Put( 1, t1.ToUniversalTime() );
Adding Multiple Entries Using PutAll
If you need to add many cache entries to a region at one time, you can improve cache performance by
using the putAll function to add them in a single distributed operation. Multiple key/value pairs are stored
in a hashmap, then the hashmap contents are processed on the server as either new entries, updates, or
invalidates for existing entries.
Under Adding an Entry to the Cache on page 90 see "Bulk Put Operations Using putAll" for more
information about the putAll API. Additional details are available in the online API documentation for
Region::putAll (C++), or Region.PutAll (.NET).
Updating Entries
A cached entry can be updated using these methods:
•
•
•
Explicitly, when a client invokes a put operation on an existing entry.
Implicitly, when a get is performed on an entry that has an invalid value in the cache. An entry can
become invalid through an explicit API call, through an automated expiration action, or by being created
with a value of null .
Automatically, when a new entry value is distributed from another cache.
Similar to entry creation, all of these operations can be aborted by a cache writer.
The get function returns a direct reference to the entry value object. A change made using that reference
is called an in-place change because it directly modifies the contents of the value in the local cache. For
details on safe cache access, see Managing the Lifetime of a Cached Object on page 82.
61
Configuring the Native Client Cache
Pivotal GemFire Native Client
Accessing Entries
Use the API to retrieve the entry key, entry value, and the RegionEntry object itself. A variety of
functions provide information for individual entries and for the set of all entries resident in the region. The
online API documentation lists all available access functions.
A region's entry keys and RegionEntry objects are directly available from the local cache. Applications
can directly access the local cache's stored entry value through the RegionEntry::getValue function.
The getValue function either returns the value if a valid value is present in the local cache, or NULL if the
value is not valid locally. This function does no data loading, nor does it look elsewhere in the distributed
system for a valid value.
Note: Direct access through RegionEntry::getValue does not reset an entry's timestamp for
LRU expiration. See Specifying Expiration Attributes on page 79 for more information about LRU
expiration.
In comparison, the standard Region::get functions consider all caches and all applicable loaders in the
distributed system in an attempt to return a valid entry value to the calling application. The primary attribute
setting affecting entry retrieval is CacheLoader. See Specifying Application Plug-In Attributes on page
74.
The standard Region::get functions may implement a number of operations in order to retrieve a valid
entry value. The operations used depend on the region's attribute settings and on the state of the entry
itself. By default, the client retrieves entry values through calls to the get function. The client can override
this behavior for any region by defining a cache loader for the region.
The following sections discuss the get function and special considerations for entry retrieval.
Entry Retrieval
You retrieve entry values with the Region::get function.
When an entry value is requested from a region, it is either retrieved from the cache server or fetched by
the region's locally-defined cache loader in this sequence:
1. local cache search
2. server cache
3. local load (For distributed regions, the local load is fetched before remote cache values)
How the get Operation Affects the Local Entry Value
If a get operation retrieves an entry value from outside the local cache through a local load, it
automatically puts the value into the cache for future reference.
Note that these load operations do not invoke a cache writer. Because the loader and writer operate
against the same data source, you do not need to perform a cache write for entries that were just fetched
from that data source. For descriptions of these processes, see Specifying Application Plug-In Attributes on
page 74.
Note: Access through a get operation resets an entry's timestamp for LRU expiration.
Getting Multiple Entries Using getAll
You can use the getAll Region API to get all values for an array of keys from the local cache or cache
server. Under Bulk Put Operations Using putAll for more information.
62
Region Consistency
Pivotal GemFire Native Client
Invalidating or Destroying Cached Entries
Invalidating an entry sets the entry's value to NULL. Destroying it removes the entry from the region
altogether. These operations can be carried out in the local cache in the following ways:
•
•
Through direct API calls from the client.
Through expiration activities based on the entry's statistics and the region's attribute settings.
Note: A user-defined cache writer is called before an operation is completed, and can abort an
entry destroy operation.
Whether carried out explicitly or through expiration activities, invalidation and destruction cause event
notification: The CacheEvent object has an isExpiration flag that is set to true for events resulting
from expiration activities, and set to false for all other events.
Notification for Operations
Listeners are invoked for client-initiated operations only after the client operation succeeds on the server.
Listener invocation on the client indicates that the server has the same data as the client.
If a client operation fails on the server, the operation is rolled back, assuming that no other thread has
modified the data in the intervening period. Rollback may not be possible in cases where the entry has
been evicted by LRU or expiration during this period. Thus when an exception is received from the server
for an operation, local changes may not have been rolled back
Event Notification Sequence
Events received on the clients that originated on the server invoke the subscription for the event as seen
by the server. Events originating on the client invoke the subscription as seen by the client.
For example, a client that receives a create and an update from the server fires a create event and
an update event because that is how the server saw it. A cacheless client that does two consecutive put
operations has two afterCreate events invoked on the originating client because the client does not
have any history about the first put , since it is cacheless.
For the same sequence, the server sees an afterCreate and an afterUpdate event, and a remote
client receiving the event sees an afterCreate followed by an afterUpdate event. A client that caches
locally sees an afterCreate and an afterUpdate for the same scenario (as will the server and remote
clients).
Region Consistency
GemFire ensures that all copies of a region eventually reach a consistent state on all members and clients
that host the region.
By default GemFire members perform consistency checks when they apply updates to a distributed region,
in order to ensure that all copies of the region eventually become consistent on all GemFire members and
client caches that host the region. Different types of region ensure consistency using different techniques.
However, when consistency checking is enabled (the default) all entries in a region require additional
overhead in order to store version and timestamp information.
Although a region must have the same consistency checking configuration on all GemFire members that
host the region, you can optionally disable consistency checking in a client cache region while leaving
consistency checking enabled for the region on GemFire members. This configuration may be necessary
63
Region Attributes
Pivotal GemFire Native Client
in certain cases where the client must view all updates to a given region, even when GemFire members
discards an update in order to preserve region consistency.
Native Client Overhead for Consistency Checks
In the native client regions, the overhead for performing consistency check is an additional 11 bytes per
region entry. This overhead is slightly smaller than the overhead required to provide consistency checking
on server-side region entries.
If you cannot support the additional overhead in your deployment, you can disable consistency checks by
setting the region attribute concurrency-checks-enabled to "false" for each region hosted by your
client.
Region Attributes
Region attributes govern the automated management of a region and its entries.
Region attribute settings determine where the data resides, how the region is managed in memory, and the
automatic loading, distribution, and expiration of region entries.
Specifying Region Attributes
Region attributes govern the automated management of a region and its entries.
Specify region attributes before creating the region. You can do this either through the declarative
XML file or through the API. The API includes classes for defining a region's attributes before creation
and for modifying some of them after creation. For details, see the online API documentation for
RegionShortcut, RegionAttributes, AttributesFactory, and AttributesMutator.
RegionShortcuts
GemFire provides a number of predefined, shortcut region attributes settings for your use, in
RegionShortcut.
Shortcut attributes are a convenience only. They are named attributes that GemFire has already stored
for you. You can override their settings by storing new attributes with the same id as the predefined
attributes.
You can also create custom region attributes and store them with an identifier for later retrieval. Both types
of stored attributes are referred to as named region attributes. You can create and store your attribute
settings in the cache.xml file and through the API.
Retrieve named attributes by providing the ID to the region creation. This example uses the shortcut
CACHING_PROXY attributes to create a region:
<region name="testRegion" refid="CACHING_PROXY"/>
You can modify named attributes as needed. For example, this adds a cache listener to the region:
<region name="testRegion" refid="CACHING_PROXY">
<region-attributes>
<cache-listener library-name="myAppLib"
library-function-name ="myCacheListener" />
</region-attributes>
</region>
64
Configuring the Native Client Cache
Pivotal GemFire Native Client
In this example, the modified region shortcut is saved to the cache using the region attribute id, for retrieval
and use by a second region:
<region name="testRegion" refid="CACHING_PROXY">
<region-attributes id="Caching_Proxy_With_Listener">
<cache-listener library-name="myAppLib"
library-function-name ="myCacheListener" />
</region-attributes>
</region>
<region name="newTestRegion" refid="Caching_Proxy_With_Listener"/>
Shortcut Attribute Options
You can select the most common region attributes settings from RegionShortcut, GemFire's predefined
named region attributes.
This section provides an overview of the options available in the region shortcut settings.
Note: For complete listings and descriptions, including information on the underlying
RegionAttributes settings for each shortcut, see the online documentation for
RegionShortcut.
These are the options available in RegionShortcut.
Communication with Servers and Data Storage
PROXY does not store data in the client cache, but connects the region to the servers for data requests
and updates, interest registrations, and so on.
CACHING_PROXY stores data in the client cache and connects the region to the servers for data
requests and updates, interest registrations, and so on.
LOCAL stores data in the client cache and does not connect the region to the servers. This is a clientside-only region.
•
•
•
Data Eviction
For the non-PROXY regions--the regions that store data in the client cache--you can add data eviction:
ENTRY_LRU causes least recently used data to be evicted from memory when the region reaches the
entry count limit.
•
Mutable and Immutable Region Attributes
Attributes that are immutable (fixed) after region creation govern storage location, data distribution,
statistics, application plug-ins, and the configuration and management of the region's data hashmap.
This table lists the immutable attributes and their default settings.
Table 12: Immutable Region Attributes
Immutable Region Attribute
Default Setting
See CachingEnabled on page 66
true
See InitialCapacity
16 (entries)
See LoadFactor
0.75
See ConcurrencyLevel
16
See DiskPolicy
See PersistenceManager
NULL
65
Configuring the Native Client Cache
Immutable Region Attribute
Pivotal GemFire Native Client
Default Setting
PartitionResolver. See Specifying Application PlugIn Attributes on page 74.
Mutable region attributes identify expiration and cache listener, cache writer and cache loader actions that
are run from the defining client. The next table lists the mutable attributes that generally can be modified
after region creation by using the AttributesMutator for the region.
Table 13: Mutable Region Attributes
Mutable Region Attribute
Default Setting
Expiration attributes. See Specifying Expiration
Attributes on page 79.
no expiration
See LruEntriesLimit on page 68.
0 (no limit)
CacheLoader. See Specifying Application Plug-In
Attributes on page 74.
CacheWriter. See Specifying Application Plug-In
Attributes on page 74.
CacheListener. See Specifying Application Plug-In
Attributes on page 74.
Under Specifying Application Plug-In Attributes on page 74,see "Using AttributesMutator to Modifty a
Plug-in" for information about using AttributesMutator with cache listeners, cache loaders, and cache
writers.
The remainder of this chapter examines these attributes in detail. Throughout the descriptions,
cache.xml file snippets show how each attribute can be set declaratively.
CachingEnabled
This attribute determines whether data is cached in this region. For example, you might choose to
configure the distributed system as a simple messaging service where clients run without a cache.
Note: You can configure the most common of these options with GemFire's predefined region
attributes. See RegionShortcuts and the Javadocs for RegionShortcut.
If CachingEnabled is false (no caching), an IllegalStateException is thrown if any of these attributes are
set:
•
•
•
•
•
•
•
InitialCapacity
EntryTimeToLive under Specifying Expiration Attributes on page 79
EntryIdleTimeout under Specifying Expiration Attributes on page 79
LoadFactor
ConcurrencyLevel
LruEntriesLimit on page 68
DiskPolicy
The following declaration enables caching for the region:
<region-attributes caching-enabled="true">
</region-attributes>
66
Configuring the Native Client Cache
Pivotal GemFire Native Client
InitialCapacity
Use this attribute, together with the LoadFactor attribute, to set the initial parameters on the underlying
hashmap that stores region entries. This is the number of entries that the region map will be ready to hold
when it is created.
This declaration sets the region's initial capacity to 10000:
<region-attributes initial-capacity="10000">
</region-attributes>
LoadFactor
Use this attribute, together with the InitialCapacity attribute, to set the initial parameters on the
underlying hashmap that stores region entries. When the number of entries in the map exceeds the
LoadFactor times current capacity, the capacity is increased and the map is rehashed. You get the best
performance if you configure a properly sized region at the start and do not have to rehash it.
This declaration sets the region's load factor to 0.75 :
<region-attributes load-factor="0.75">
</region-attributes>
ConcurrencyLevel
This attribute estimates the maximum number of application threads that concurrently access a region
entry at one time. This attribute helps optimize the use of system resources and reduce thread contention.
The following declaration sets the region's ConcurrencyLevel to 16:
<region-attributes concurrency-level="16">
</region-attributes>
Note: When CachingEnabled is false, do not set the ConcurrencyLevel attribute. An
IllegalStateException is thrown if the attribute is set.
ConcurrencyChecksEnabled
This attribute determines whether members perform checks to provide consistent handling for concurrent
or out-of-order updates to distributed regions.
A client cache can disable consistency checking for a region even if server caches enable consistency
checking for the same region. This configuration ensures that the client sees all events for the region, but it
does not prevent the client cache region from becoming out-of-sync with the server cache.
Optionally enable concurrency checks for the region. Example:
<region-attributes concurrency-checks-enabled="true">
</region-attributes>
See Region Consistency on page 63 for more information.
67
Configuring the Native Client Cache
Pivotal GemFire Native Client
LruEntriesLimit
This attribute sets the maximum number of entries to hold in a caching region. When the capacity of the
caching region is exceeded, a least-recently-used (LRU) algorithm is used to evict entries.
Note: This is a tuning parameter that affects system performance.
When eviction is configured, memory consumption or entry count is monitored and, when capacity is
reached, GemFire makes way for new entries by removing or overflowing the stalest LRU entries to disk.
If you use disk data overflow to supplement memory for your data cache, make sure you have enough disk
space to store the data.
This declaration limits the region to 20,000 entries:
<region-attributes lru-entries-limit="20000"
initial-capacity="20000"
load-factor="1">
</region-attributes>
Evicted entries can be destroyed or moved to disk as an extension of the cache. See DiskPolicy.
Note: When CachingEnabled is false, do not set the LruEntriesLimit attribute. An
IllegalStateException is thrown if the attribute is set.
See also Controlling Cache Size.
DiskPolicy
If the lru-entries-limit attribute is greater than zero, the optional disk-policy attribute determines
how over-limit LRU entries are handled. LRU entries over the limit are either destroyed by default (diskpolicy is none ) or written to disk (overflows).
Note: If LruEntriesLimit is 0, or CachingEnabled is false, do not set the disk-policy
attribute. An IllegalStateException is thrown if the attribute is set.
This declaration causes LRU to overflow to disk:
<region-attributes lru-entries-limit="20000"
disk-policy="overflows">
<persistence-manager ... />
</region-attributes>
Overflow requires a persistence manager for cache-to-disk and disk-to-cache operations. See
PersistenceManager.
Overflowing Data to Disk
Region data can be stored to disk using the overflow process to satisfy region capacity restrictions without
completely destroying the local cache data. The storage mechanism uses disk files to hold region entry
data. When an entry is overflowed, its value is written to disk but its key and entry object remain in the
cache. This also uses the region attribute DiskPolicy.
Overflow allows you to keep the region within a user-specified size in memory by relegating the values of
least recently used (LRU) entries to disk. Overflow essentially uses disk as a swap space for entry values.
When the region size reaches the specified threshold, entry values are moved from memory to disk, as
68
Configuring the Native Client Cache
Pivotal GemFire Native Client
shown in the following figure. If an entry is requested whose value is only on disk, the value is copied back
into memory, possibly causing the value of a different LRU entry to be overflowed to disk.
Figure 2: Data Flow Between Overflow Region and Disk Files
In this figure the value of the LRU entry X has been moved to disk to recover space in memory. The key for
the entry remains in memory. From the distributed system perspective, the value on disk is as much a part
of the region as the data in memory. A get performed on region B looks first in memory and then on disk
as part of the local cache search.
PersistenceManager
For each region, if the disk-policy attribute is set to overflows, a persistence-manager plug-in must perform
cache-to-disk and disk-to-cache operations. See Application Plug-Ins.
Persistence manager declaration:
<region-attributes lru-entries-limit="nnnnn"
disk-policy="overflows">
<persistence-manager library-name="libraryName"
library-function-name="functionName">
<properties>
<property name="propertyName" value="propertyValue" />
</properties>
</persistence-manager>
</region-attributes>
The optional properties set parameters for the plug-in.
•
•
•
•
•
Using SQLite as a Persistence Manager on page 69
SQLite Persistence Manager Region Attributes on page 70
Configuring the SQLite Persistence Manager Plug-In for C++ Applications on page 70
Configuring the SQLite Persistence Manager Plug-In for .NET Applications on page 71
Implementing a PersistenceManager with the IPersistenceManager Interface on page 72
Using SQLite as a Persistence Manager
The GemFire native client distribution includes a persistence manager that uses the open-source SQLite
library.
SQLite is a software library that implements a self-contained transactional SQL database. SQLite does not
require its own server or separate configuration, and the source code for SQLite is in the public domain.
For more information on SQLite, see http://www.sqlite.org.
Each SQLite persistence manager persists its region data in a SQLite database that is stored in disk
files. In a given native client application process, each region must have a unique persistence (overflow)
directory.
69
Configuring the Native Client Cache
Pivotal GemFire Native Client
Figure 3: SQLite Database Persistence Manager Directory Structure
SQLite Persistence Manager Region Attributes
The following table describes the region attributes that can be configured for the SQLite persistence
manager.
Property
Description
Default Setting
PersistenceDirectory
Directory where each region's database files
are stored. This setting must be different for
each region including regions in different
processes. This directory is created by
the persistence manager. The persistence
manager fails to initialize if this directory
already exists or cannot be created.
Default is to create a
subdirectory named
GemFireRegionData in the
directory where the process
using the region was started.
PageSize
Maximum page size of the SQLite database.
SQLite can limit the size of a database file to
prevent the database file from growing too
large and consuming too much disk space.
Ordinarily, if no value
is explicitly provided,
SQLite creates a database
with the page size set to
SQLITE_DEFAULT_PAGE_SIZE
(default is 1024). However,
based on certain device
characteristics (for example,
sector-size and atomic write()
support) SQLite may choose a
larger value. PageSize specifies
the maximum value that SQLite
will be able to choose on its
own. See http://www.sqlite.org/
compile.html#default_page_size
for more details on
SQLITE_DEFAULT_PAGE_SIZE.
MaxPageCount
Maximum number of pages in one database
file.
SQLite default, which is
1073741823.
Configuring the SQLite Persistence Manager Plug-In for C++ Applications
To load the SQLite persistence manager plug-in for C++ applications, you can configure it either in your
client's cache.xml or programmatically using the native client C++ API.
The following is an example of how to specify the following region attributes in your client's cache.xml:
<region-attributes>
70
Configuring the Native Client Cache
Pivotal GemFire Native Client
<persistence-manager library-name="libSqLiteImpl.so" library-functionname="createSqLiteInstance">
<properties>
<property name="PersistenceDirectory" value="/xyz"/>
<property name="PageSize" value="65536"/>
<property name="MaxPageCount" value="1073741823"/>
</properties>
</persistence-manager>
</region-attributes>
C++ API Example
To use the native client C++ API, set SQLite persistence manager attributes
programmatically as follows:
PropertiesPtr sqliteProperties = Properties::create();
sqliteProperties->insert("MaxPagecount""5");
sqliteProperties->insert("PageSize","1024");
sqliteProperties->insert("PersistenceDirectory","SqLiteTest779");
regionFactory>setPersistenceManager("SqLiteImpl","createSqLiteInstance",sqliteProperties);
Configuring the SQLite Persistence Manager Plug-In for .NET Applications
To load the SQLite persistence manager plug-in for .NET applications, you can configure it either in your
client's cache.xml or programmatically using the .NET API:
<persistence-manager library-name="Gemstone.Gemfire.Plugins.SqLite"
library-functionname="GemStone.GemFire.Plugins.SqLite.SqLiteImpl&lt;System.Object,
System.Object&gt;.Create">
<properties>
<property name="PersistenceDirectory" value="SqLite"/>
<property name="MaxPageCount" value="1073741823"/>
<property name="PageSize" value="65536"/>
</properties>
</persistence-manager>
.NET API Example
To use the native client .NET API, set the SQLite persistence manager attributes
programmatically as follows:
Properties<string, string> sqliteProperties = new
Properties<string, string>();
sqliteProperties.Insert("PageSize", "65536");
sqliteProperties.Insert("MaxFileSize", "51200000");
sqliteProperties.Insert("PersistenceDirectory", SqLiteDir);
rf.SetPersistenceManager("Gemstone.Gemfire.Plugins.SqLite",
"Gemstone.Gemfire.Plugins.SqLiteSqLiteImpl<System.Object,System.Object>.Create",
sqliteProperties);
You can also use and configure the C++ SQLite persistence manager library from
your .NET application as follows:
rf.SetPersistenceManager("SqliteImpl", "createSqLiteInstance",
sqliteProperties);
71
Configuring the Native Client Cache
Pivotal GemFire Native Client
Implementing a PersistenceManager with the IPersistenceManager Interface
When developing .NET managed applications, you can use the IPersistenceManager managed interface to
implement your own persistence manager. The following code sample provides the IPersistenceManager
interface:
/// <summary>
/// IPersistenceManager interface for persistence and overflow.
/// This class abstracts the disk-related operations in case of persistence
or overflow to disk.
/// A specific disk storage implementation will implement all the methods
described here.
/// </summary>
generic<class TKey, class TValue>
public interface class IPersistenceManager
{
public:
/// <summary>
/// Called after an implementation object is created. Initializes all the
implementation specific environments needed.
/// </summary>
/// <param name="region">
/// Region for which this PersistenceManager is initialized.
/// </param>
/// <param name="diskProperties">
/// Configuration Properties used by PersistenceManager implementation.
/// </param>
void Init(IRegion<TKey, TValue>^ region, Properties<String^, String^>^
diskProperties);
/// <summary>
/// Writes a key, value pair of region to the disk. The actual file or
database related write operations should be implemented in this method.
/// </summary>
/// <param name="key">
/// the key to write.
/// </param>
/// <param name="value">
/// the value to write.
/// </param>
void Write(TKey key, TValue value);
/// <summary>
/// This method is not used.
/// </summary>
bool WriteAll();
/// <summary>
/// Reads the value for the key from the disk.
/// </summary>
/// <param name="key">
/// key for which the value has to be read.
/// </param>
TValue Read(TKey key);
/// <summary>
/// This method is not used.
/// </summary>
bool ReadAll();
/// <summary>
/// Destroys the entry specified by the key in the argument.
/// </summary>
72
Configuring the Native Client Cache
Pivotal GemFire Native Client
/// <param name="key">
/// key of the entry which is being destroyed.
/// </param>
void Destroy(TKey key);
}
/// <summary>
/// Closes the persistence manager instance.
/// </summary>
void Close();
The following is a sample interface implementation:
class MyPersistenceManager<TKey, TValue> : IPersistenceManager<TKey,
TValue>
{
#region IPersistenceManager<TKey,TValue> Members
public void Close()
{
throw new NotImplementedException();
}
public void Destroy(TKey key)
{
throw new NotImplementedException();
}
public void Init(IRegion<TKey, TValue> region, Properties<string,
string> disk Properties)
{
throw new NotImplementedException();
}
public TValue Read(TKey key)
{
throw new NotImplementedException();
}
public void Write(TKey key, TValue value)
{
throw new NotImplementedException();
}
public bool ReadAll()
{
throw new NotImplementedException();
}
}
public bool WriteAll()
{
throw new NotImplementedException();
}
#endregion
73
Configuring the Native Client Cache
Pivotal GemFire Native Client
Specifying Application Plug-In Attributes
The plug-in attributes allow you to customize client region behavior for loading, updating, deleting, and
overflowing region data and for accessing data in server partitioned regions. All client plug-ins are available
through the C++ and .NET API.
Application plug-ins for cache regions in clients can be declared either programmatically or in the
cache.xml file.
•
•
•
•
•
•
•
•
Figure 4: Where Application Plug-Ins Run on page 74
Overview of Application Plug-ins on page 74
CacheLoader on page 75
CacheWriter on page 75
CacheListener on page 75
PartitionResolver on page 76
Using AttributesMutator to Modify a Plug-In on page 78
Considerations for Implementing Callbacks on page 79
Figure 4: Where Application Plug-Ins Run
Overview of Application Plug-ins
The API provides the framework for application plug-ins with callback functions for the appropriate events.
Your classes and functions can customize these for your application's needs. When creating a region,
specify these as part of the region's attributes settings. For regions already in the cache, you can specify
new CacheLoader, CacheWriter, and CacheListener using the region's AttributesMutator. The
PartitionResolver is not mutable.
•
•
•
•
CacheLoader: A data loader called when an entry get operation fails to find a value for a given key.
A cache loader is generally used to retrieve data from an outside source such as a database, but it
may perform any operation defined by the user. Loaders are invoked as part of the distributed loading
activities for entry retrieval, described in Entry Retrieval.
CacheWriter: A synchronous event listener that receives callbacks before region events occur and
has the ability to abort the operations. Writers are generally used to keep a back-end data source
synchronized with the cache.
CacheListener: An asynchronous event listener for region events in the local cache.
PartitionResolver: Used for single-hop access to partitioned region entries on the server side. This
resolver implementation must match that of the PartitionResolver on the server side.
The following XML declaration specifies a cache loader for a region when the region is created.
<region-attributes>
<cache-loader library-name="appl-lib"
library-function-name ="createCacheLoader">
74
Configuring the Native Client Cache
Pivotal GemFire Native Client
</cache-loader>
</region-attributes>
The rest of this section gives more detailed descriptions of these application plug-ins, followed by special
considerations for plug-ins in distributed regions and some guidelines for writing callbacks.
CacheLoader
A cache loader is an application plug-in used to load data into the region. When an entry is requested
that is unavailable in the region, a cache loader may be called upon to load it. Generally, you use a cache
loader to retrieve the data from a database or another source outside the distributed system, but it may
perform any operation defined by the user.
The CacheLoader interface provides one function, load, for customizing region entry loading. A
distributed region may have cache loaders defined in any or all caches where the region is defined. When
loading an entry value, a locally defined cache loader is always used before a remote loader. In distributed
regions, loaders are available for remote entry retrieval.
CacheWriter
A cache writer is an application plug-in that synchronously handles changes to a region's contents.
It is generally used to keep back-end data sources synchronized with a cache region. A cache writer
has callback functions to handle region destruction and entry creation, update, and destruction. These
functions are all called before the modification has taken place and can abort the operation.
You can also use cache writers to store data that you want to make persistent.
CacheListener
A cache listener is an application plug-in that asynchronously handles changes to a region's contents.
A cache listener has callback functions to handle region destruction and invalidation, along with entry
creation, update, invalidation, and destruction. These functions are called asynchronously after the
modification has taken place.
This declarative XML example establishes a cache listener when a region is created:
<region name="region11">
<region-attributes>
<cache-listener library-name="appl-lib"
library-function-name ="createCacheListener" />
</region-attributes>
</region>
Unlike cache loaders and cache writers, cache listeners only receive events for entries to which the client
has performed operations or registered interest.
When the listener is attached to a region with caching disabled, the old value is always NULL.
Note: Do not perform region operations inside the cache listener. Once you have configured a
cache listener, the event supplies the new entry values to the application. Performing a get with a
key from the EntryEvent can result in distributed deadlock. For more about this, see the online
API documentation for EntryEvent.
When a region disconnects from a cache listener, you can implement the afterRegionDisconnected
callback event. This callback event is only be invoked when using the pool API and subscription is
enabled on the pool. For example:
class DisconnectCacheListener : public CacheListener
{
void afterRegionDisconnected( const RegionPtr& region )
{
75
Configuring the Native Client Cache
};
}
Pivotal GemFire Native Client
printf("After Region Disconnected event received");
PartitionResolver
This section pertains to data access in server regions that have custom partitioning. Custom partitioning
uses a Java PartitionResolver to colocate like data in the same buckets. For the client, you can
use a PartitionResolver that matches the server's implementation to access data in a single hop.
With single-hop data access, the client pool maintains information on where a partitioned region's data is
hosted. When accessing a single entry, the client directly contacts the server that hosts the key--in a single
hop.
Note: Single hop is only used for these operations that are run on single data points: put, get,
getAll, destroy.
Implementing Single-Hop on a Partitioned Region
1. Make sure the pool attribute, pr-single-hop-enabled, is set to true or not set. It is true by
default.
2. If the server uses a custom PartitionResolver install an implementation of PartitionResolver
in the client region that returns, entry for entry, the same value as the server's Java
PartitionResolver implementation. The server uses the resolver to colocate like data within a
partitioned region.
If the server does not use a custom resolver, the default resolvers in client and server match, so single
hop will work there by default.
Disable single hop behavior for a region by setting its pool attribute pr-single-hop-enabled to false.
See <pool> in the GemFire User's Guide for information on setting pr-single-hop-enabled.
See the GemFire User's Guide for more information, including co-locating like data within a partitioned
region and how to get the best performance with PR single hop.
Implementing a PartitionResolver
See the GemFire User's Guide for information on custom-partitioning the server partitioned regions.
1. Implement PartitionResolver in the same place that you did in the server--custom class, key, or
cache callback argument.
2. Program the resolver's functions the same way you programmed them in the Java implementation. Your
implementation must match the server's.
Example of programming the PartitionResolver in C++:
class TradeKeyResolver : public PartitionResolver
{
private:
string m_tradeID;
int m_month;
int m_year;
public:
TradeKeyResolver() { }
TradeKeyResolver(int month, int year) {
m_month = month;
m_year = year;
}
~TradeKeyResolver() { }
static PartitionResolverPtr createTradeKeyResolver() {
PartitionResolverPtr tradeKeyResolver( new TradeKeyResolver());
76
Configuring the Native Client Cache
};
Pivotal GemFire Native Client
return tradeKeyResolver;
}
const char* getName() {
return "TradeKey";
}
CacheableKeyPtr getRoutingObject(const EntryEvent& opDetails) {
return CacheableKey::create(m_month + m_year);
}
Example of programming the PartitionResolver in C#:
using System;
using System.Threading;
// Use the GemFire namespace
using GemStone.GemFire.Cache;
class TradeKeyResolver : IPartitionResolver
{
private int m_month = 0;
private int m_year = 0;
public static TradeKeyResolver CreateTradeKeyResolver()
{
return new TradeKeyResolver();
}
public virtual ICacheableKey GetRoutingObject(EntryEvent entry)
{
return new CacheableInt32(m_month + m_year);
}
}
public virtual String GetName()
{
return "TradeKeyResolver";
}
3. Install the resolver in the region. Use one of these methods:
XML partition resolver declaration:
<region name="trades" refid="CACHING_PROXY">
<region-attributes>
<partition-resolver library-name="appl-lib" library-function-name=
"createTradeKeyResolver"/>
</region-attributes>
</region>
<pool free-connection-timeout="12345" idle-timeout="5555"
load-conditioning-interval="23456" max-connections="7"
min-connections="3" name="test_pool_1" ping-interval="12345"
read-timeout="23456" retry-attempts="3" servergroup="ServerGroup1"
socket-buffer-size="32768" statistic-interval="10123"
subscription-ack-interval="567" subscription-enabled="true"
subscription-message-tracking-timeout="900123"
subscription-redundancy="0" thread-local-connections="5"
pr-single-hop-enabled="true" >
<locator host="localhost" port="34756"/>
</pool>
Programmatic partition resolver installation:
void setPartitionResolver()
77
Configuring the Native Client Cache
{
}
Pivotal GemFire Native Client
CachePtr cachePtr = CacheFactory::createCacheFactory()->create();
PartitionResolverPtr resolver( new TradeKeyResolver());
RegionFactoryPtr regionFactory =
cachePtr->createRegionFactory(PROXY)
->setClientNotificationEnabled(true)
->setPartitionResolver(resolver);
RegionPtr regionPtr = regionFactory->create( "Trades" );
Your implementation of PartitionResolver must match that of the server side.
Using AttributesMutator to Modify a Plug-In
A cache listener, cache loader or cache writer can be added to or removed from a region after the region
is created by retrieving and running the Region object's AttributesMutator. Mutable attributes define
operations that are run from the client itself.
This example shows how to use AttributesMutator to dynamically add a cache listener to an existing
region.
void setListener(RegionPtr& region)
{
CacheListenerPtr regionListener = new TestCacheListener();
AttributesMutatorPtr regionAttributesMutator =
region->getAttributesMutator();
}
// Change cache listener for region.
regionAttributesMutator->setCacheListener(regionListener);
The plug-ins can also be implemented using a dynamically linked library. The class is not available to the
application code in this case, so a factory method is required by the set function along with the name of
the library.
This example shows how to use AttributesMutator along with the setCacheListener function to
obtain a new cache listener object using the factory function provided by the library. Next, the listener is
set for the region.
void setListenerUsingFactory(RegionPtr& region)
{
AttributesMutatorPtr regionAttributesMutator =
region->getAttributesMutator();
// Change cache listener for region.
regionAttributesMutator->setCacheListener("Library",
"createTestCacheListener");
}
To use AttributesMutator to remove a plug-in from a region, set the plug-in's value to NULLPTR, as
shown in the following example.
void removeListener(RegionPtr& region)
{
CacheListenerPtr nullListener = NULLPTR;
AttributesMutatorPtr regionAttributesMutator =
region->getAttributesMutator();
}
// Change cache listener for region to NULLPTR
regionAttributesMutator->setCacheListener(nullListener);
78
Configuring the Native Client Cache
Pivotal GemFire Native Client
Considerations for Implementing Callbacks
Keep your callback implementations lightweight and prevent situations that might cause them to hang. For
example, do not perform distribution operations or disconnects inside events.
Your code should handle any exceptions that it generates. If not, GemFire handles them as well as
possible. Because C++ has no standard for exceptions, in many cases GemFire can only print an
unknown error message.
Specifying Expiration Attributes
Expiration attributes govern the automatic eviction of regions and region entries from the cache. Eviction
is based on the time elapsed since the last update or access to the object. This is referred to as the leastrecently-used (LRU) eviction process. Expiration options range from marking the expired object as invalid
to completely removing it from the distributed cache. Eviction can help keep data current by removing
outdated entries, prompting a reload the next time they are requested. Eviction may also be used to
recover space in the cache by clearing out unaccessed entries and regions.
Similar to application plug-ins, expiration activities are hosted by each application that defines a region in
its cache.
The following example shows a declaration that causes the region's entries to be invalidated in the local
cache after they have not been accessed for one minute.
<region-attributes>
<entry-idle-time>
<expiration-attributes timeout="60" action="local-invalidate"/>
</entry-idle-time>
</region-attributes>
Region and region entry expiration attributes are set at the region level. By default, regions and entries do
not expire. The following attributes cover two types of expiration: time-to-live (TTL) and idle timeout.
RegionTimeToLive
Number of seconds that the region remains in the
cache after the last creation or update before the
expiration action occurs.
EntryTimeToLive
For entries, the counter is set to zero for create
and put operations. Region counters are reset
when the region is created and when an entry has
its counter reset. An update to an entry causes the
time-to-live (TTL) counters to be reset for the entry
and its region.
RegionIdleTimeout
Number of seconds that the region remains in the
cache after the last access before the expiration
action occurs.
EntryIdleTimeout
The idle timeout counter for an object is reset
when its TTL counter is reset. An entry's idle
timeout counter is also reset whenever the entry is
accessed through a get operation.
The idle timeout counter for a region is reset
whenever the idle timeout is reset for one of its
entries.
79
Cache Management
Pivotal GemFire Native Client
Using Statistics to Measure Expiration
Expiration is measured by comparing expiration attribute settings with the last accessed time
and last modified time statistics. These statistics are directly available to applications through
the CacheStatistics object that is returned by the Region::getStatistics and
RegionEntry::getStatistics functions. The CacheStatistics object also provides a function for
resetting the statistics counters.
Expiration Actions
You can specify one of the following actions for each expiration attribute:
•
•
•
•
Destroy. Removes the object completely from the cache. For regions, all entries are destroyed as well.
Destroy actions are distributed according to the region's distribution settings.
Invalidate. Marks the object as invalid. For entries, the value is set to NULL. You invalidate a region
by invalidating all its entries. Invalidate actions are distributed according to the region's distribution
settings. When an entry is invalid, a get causes the cache to retrieve the entry according to the steps
described in Entry Retrieval.
Local destroy. Destroys the object in the local cache but does not distribute the operation.
Local invalidate. Invalidates the object in the local cache but does not distribute the operation.
Note: Destruction and invalidation cause the same event notification activities whether carried
out explicitly or through expiration activities.
Region Expiration
Expiration activities in distributed regions can be distributed or performed only in the local cache. So one
cache could control region expiration for a number of caches in the distributed system.
Cache Management
This chapter covers cache management.
Client-to-Server Connection Process
It is important to understand the sequence of events that occur when the native client connects with a
GemFire cache server.
1. A native client region is configured in cache.xml or programmatically with a set of server connection
endpoints. Server endpoints identify each cache server by specifying the server's name and port
number.
Client threads obtain, use, and release a connection to a connection pool that maintains
new connections. The number of connections that a client can establish is governed by
the pool's min-connections and max-connections settings, either using client XML
configuration or programmatically through the CacheFactory::setMinConnections() and
CacheFactory::setMaxConnections() APIs. The defaults for min-connections is 1 and maxconnections is -1 meaning the connection count can grow to accommodate the number of active
threads performing region operations.
This example shows how to use cache.xml to configure a native client region with endpoints set to
two cache servers:
<pool name="examplePool" subscription-enabled="true" >
<server host="java_servername1" port="java_port1" />
80
Configuring the Native Client Cache
Pivotal GemFire Native Client
<server host="java_servername2" port="java_port2" />
</pool>
<region name="NativeClientRegion" refid="CACHING_PROXY">
<region-attributes pool-name="examplePool"/>
</region>
TCP connections on the native client are specified at the cache level, or by overriding endpoints for
specific regions. The connections are created as the regions are created. In addition, connections can
also get created for querying without having any created regions. In this case, when endpoints are
defined at the cache level no regions are yet created and a query is fired.
You can configure client-server connections in two ways. Use either the region/cache endpoints or the
Pool API. For more information about the pool API, see Using Connection Pools.
2. The client announces to the server which entries it wishes to have updated by programmatically
registering interest in those entries. See Registering Interest for Entries for more information.
3. The client cache.xml file should have the following parameters configured so the client can update the
server and the client can receive updates from the server:
Caching enabled in the client region, by using the CACHING_PROXY RegionShortcut setting in
the region attribute refid. A listener could also be defined so event notification occurs. You can use
both, but at least one of the two methods must be used by the client to receive event notifications.
• Set subscription-enabled to true so the client receives update notifications from the server for
entries to which it has registered interest.
4. A native client application calls the C++ or .NET API to connect to a cache server.
5. The client and the cache server exchange a handshake over a configured endpoint to create a
connection.
6. Any create, put, invalidate, and destroy events sent to the server are propagated across the
distributed cache so the client can receive the events.
•
Note: You may be able to improve system performance by making adjustments to the cache
server. For cache server tuning information, see the GemFire User's Guide.
Controlling Cache Size
You can control cache size through region size limits, cache size limits, or a combination of the two.
GemFire controls region size by moving least recently used (LRU) entries from the region or from all cache
regions.
Controlling Region Size
You can cap the size of any region with the region attribute LruEntriesLimit on page 68. You can specify
whether to destroy the entries or overflow them to disk in the attribute DiskPolicy. If you overflow entries to
disk, you must also specify the attribute PersistenceManager.
Controlling Cache Size
You can control overall cache size with the heap-lru-limit, which is set in gfcpp.properties. This
property sets the maximum amount of memory used for the cache, in megabytes. If a new entry causes
memory to grow past this limit, GemFire's LRU algorithm is called to evict entries. Heap LRU causes
eviction to occur on all regions in the cache, overriding region-level LruEntriesLimit on page 68 settings
when it needs to reclaim memory.
For each region, evictions are performed according to the region's DiskPolicy and
PersistenceManager settings. If you use heap-lru-limit, review these region attributes for all your
caching regions, to be sure you are evicting the way you want to.
81
Configuring the Native Client Cache
Pivotal GemFire Native Client
The related heap-lru-delta property, also set in gfcpp.properties, is the amount of memory to free
up once the LRU evictions have begun. Memory is reclaimed until the amount of memory used is below
heap-lru-limit minus heap-lru-delta.
Managing the Lifetime of a Cached Object
All cacheable objects derive from SharedBase , which provides reference counting. Cacheable objects
are referenced using SharedPtr types.
When SharedPtr retrieves a cached object, the object remains alive as long as that pointer or the cache
itself references the object.
A native client may have many pointers that reference an object. Regardless of how many pointers to the
object are deleted, the object remains alive until the last remaining pointer is deleted. At that point the
object is deleted.
This is a very simple example:
CacheableStringPtr p = CacheableString::create ("string");
region.put ("key", p) ;
In the example:
•
•
•
The act of object creation allocates memory and initializes the object.
When you assign the object to a SharedPtr , you relinquish control of the lifetime of that object to the
reference counting mechanism for the cache.
The put operation does not actually copy the object into the cache. Rather, it copies a SharedPtr
into the cache's hashmap. Consequently, the object remains alive in the cache when the original
SharedPtr goes away.
The client can make use of an object after you have initialized the object. For example, another
SharedPtr might issue a get to retrieve the object from the cache:
CacheableStringPtr p2 = region.get ("key") ;
Because p (the original SharedPtr ) and p2 point to the same object in memory, it is possible under
some circumstances for multiple SharedPtr types to work on the same object in data storage.
Note: Once you have put an object into the cache, do not delete it explicitly. Attempting to do so
can produce undesirable results.
Changed Objects
If an object update is received, the cache no longer holds the same object. Rather, it holds a completely
different instance of the object. The client does not see the updates until it calls a get to fetch the object
again from the local cache, or (in a cache plug-in) calls EntryEvent::getNewValue.
For more about plug-ins, see Application Plug-Ins.
Object Expiration
When a cache automatically deletes an object as a result of an expiration action, the reference counting
pointers protect the client from situations that might otherwise result if the cache actually freed the object's
memory. Instead, the client disconnects the object from the cache by deleting the cache's SharedPtr
reference, while leaving untouched any client threads with a SharedPtr to that object.
82
Configuring the Native Client Cache
Pivotal GemFire Native Client
Object Lifetime Across the Distributed Cache
An object remains alive until every copy of the object is gone. In distributed regions, expiration activities
can be local or distributed, depending on a region's distribution settings. One cache could control the
expiration of all copies of an object in all the caches in the distributed system. Alternatively, each cache
could control the expiration of its own local copy of the object. If the configuration gives each cache local
control, and the expiration parameters are set to different lengths of time in different caches, some copies
of an object may still exist after it has disappeared in other caches. See Expiration Attributes for more
information.
Using Thread Safety in Cache Management
When you perform structural changes on your cache, such as creating or closing a Cache, Pool, or
Region, synchronize your operations or do them in a single thread.
Other non-structural operations, like region gets, puts, and queries, are thread safe, and you can perform
them in a multithreaded way. There are caveats to this, for example, when two threads update the same
key simultaneously, there is no way to determine which thread's operation will prevail.
You may need to protect cached objects from concurrent usage and modification. The native client does
not guard cached objects themselves from concurrent access.
Always catch and handle exceptions that may be thrown, for problems like trying to create a Pool with the
same name more than once.
Troubleshooting
This section provides troubleshooting information for the native client.
•
•
Cannot Acquire Windows Performance Data on page 83
Generating a Process Memory Dump Image for Fatal Errors on page 84
Cannot Acquire Windows Performance Data
When you attempt to run performance measurements for the native client on Windows, you may encounter
the following error message in the run logs:
Can't get Windows performance data. RegQueryValueEx returned 5
This can occur because incorrect information is returned when a Win32 application calls the ANSI version
of RegQueryValueEx Win32 API with HKEY_PERFORMANCE_DATA. This error is described in Microsoft
KB article ID 226371 at http://support.microsoft.com/kb/226371/en-us.
To successfully acquire Windows performance data, you need to verify that you have the proper registry
key access permissions in the system registry. In particular, make sure that Perflib in the following
registry path is readable (KEY_READ access) by the GemFire process:
HKEY_LOCAL_MACHINE\
SOFTWARE\
Microsoft\
Windows NT\
CurrentVersion\
Perflib
An example of reasonable security on the performance data would be to grant administrators
KEY_ALL_ACCESS and interactive users KEY_READ access. This particular configuration prevents nonadministrator remote users from querying performance data.
83
Configuring the Native Client Cache
Pivotal GemFire Native Client
See http://support.microsoft.com/kb/310426 and http://support.microsoft.com/kb/146906 for instructions
about how to ensure that GemFire processes have access to the registry keys associated with
performance.
Generating a Process Memory Dump Image for Fatal Errors
You can generate a process memory dump image (core files in Unix systems and minidumps in Windows).
The image is produced when a fatal error occurs that normally terminates the program.
When the system property crash-dump-enabled is set to true, a dump image is generated (the default
is true). The dump file is generated in the same location as the log-file directory, and has the same
prefix as the log file. The name is <prefix>-<time>.core.<pid> in Unix, and <prefix>-<time><pid>.dmp in Windows).
Unix systems generate core files automatically for such errors, but this option is useful for providing a
custom location and name, as well as for systems where core dump generation is disabled. For Unix, when
system core dump generation is turned on (ulimit -c) this property can be set to false.
For .NET clients, when this property is set then AccessViolation exceptions are trapped and a crash
dump is created to assist with further analysis. Applications receive a FatalInternalException for this
case, with the InnerException set to the originating AccessViolationException.
This requires the availability of dbghelp.dll on Windows, either in the same directory as
gfcppcache.dll or in the system PATH. The file is installed by default, though for Windows 2000 a
newer version may be required for minidumps. For Unix systems, the gcore command should be available
(gdb > 5.2 on Linux; available by default in Solaris).
84
Configuring the Native Client Cache
Pivotal GemFire Native Client
85
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
Part
VI
Pivotal GemFire Native Client C++ API
Topics:
•
•
•
•
•
•
•
•
•
About the Native Client C++ API
Creating a Cache
Creating a Proxy Client-Side
Region
Adding an Entry to the Cache
Accessing an Entry
Serializing Data
Implementing User-Defined
Objects in Java Clients
Using a Custom Class
Creating New Statistics
Pivotal GemFire Native Client C++ API describes the primary classes,
and usage conventions for the native client C++ API. It demonstrates
how to use the API to create caches and perform data serialization.
The native client C++ API allows C++ and .NET developers to
programmatically create, populate, and manage a Pivotal GemFire
distributed system. The C++ library is thread-safe, except where
specified otherwise in the online API documentation.
The online C++ API documentation is included in the docs/cppdocs
directory of the GemFire native client product installation). It provides
extensive implementation details for the C++ structures and functions.
Several example API programs are included in the examples
directory. See QuickStart Examples and Guide on page 33.
86
About the Native Client C++ API
Pivotal GemFire Native Client
About the Native Client C++ API
The native client C++ API allows C++ and .NET developers to programmatically create, populate, and
manage a Pivotal GemFire distributed system. The C++ library is thread-safe, except where specified
otherwise in the online API documentation.
This chapter gives a general overview of the classes in the gemfire, gemfire_statistics, and
gemfire_admin namespaces. For complete and current information on the classes listed here, see the
online C++ API documentation.
•
•
•
•
•
•
Cache Classes on page 87
Region Classes on page 87
Region Attribute Classes on page 87
Application Plug-In Classes on page 87
Event Handling Classes on page 88
Statistics API on page 88
Cache Classes
The native client C++ API has the following cache classes:
•
•
CacheFactory. Use this class to create and configure a Cache instance. If cache.xml is specified, the
cache is created based on the declarations loaded from that file.
Cache. Entry point to the client caching API. The cache is created by calling the create
function of the factory class, CacheFactory. Regions are configured and obtained using the
Cache::createRegionFactory() API.
Region Classes
The native client C++ API has the following region classes:
•
Region. Provides functions for managing regions and cached data. Use these functions to perform the
following actions:
•
•
•
•
•
Retrieve information about the region, such as its parent region and region attribute objects.
Invalidate or destroy the region.
Create, update, invalidate and destroy region entries.
Retrieve region entry keys, entry values, and RegionEntry objects, either individually or as entire
sets.
• Retrieve the statistics object associated with the region.
• Set and get user-defined attributes.
RegionEntry. Contains the key and value for the entry, and provides all non-distributed entry
operations. This object's operations are not distributed and do not affect statistics.
Region Attribute Classes
The native client C++ API has the following region attribute classes:
•
•
RegionAttributes. Holds all attribute values for a region and provides functions for retrieving all
attribute settings. This class can be modified by the AttributesMutator class after region creation.
AttributesMutator . Allows modification of an existing region’s attributes for application plug-ins and
expiration actions. Each region has an AttributesMutator instance.
Application Plug-In Classes
The native client C++ API has the following application plug-in classes:
87
Creating a Cache
•
•
•
Pivotal GemFire Native Client
CacheLoader. Loads data into a region on a cache miss.
CacheWriter. Synchronously handles region and entry events before the events occur. Entry events
are create, update, invalidate, and destroy. Region events are invalidate and destroy. This
class has the ability to abort events.
CacheListener. Handles region and entry events after they occur. Entry events are create, update,
invalidate, and destroy. Region events are invalidate and destroy.
Event Handling Classes
The native client C++ API has the following event handling classes:
•
•
RegionEvent. Provides information about the event, such as what region the event originated in,
whether the event originated in a cache remote to the event handler, and whether the event resulted
from a distributed operation.
EntryEvent. Provides all available information for the RegionEvent, and provides entry-specific
information such as the old and new entry values and whether the event resulted from a load
operation.
Statistics API
The StatisticsType API represents a blueprint for the same type of Statistics.
The StatisticsType API is a collection of StatisticDescriptor. Internally, each
StatisticDescriptor describes data of each individual statistic. StatisticsFactory provides
functionality for creating StatisticDescriptor, StatisticsType, and Statistics.
•
•
•
•
•
CacheStatistics --This class defines common statistics functions. Region and RegionEntry both
have functions that return a CacheStatistics object for accessing and resetting their statistics
counts.
StatisticDescriptor. An instance of this class describes a statistic whose value is updated by an
application and may be archived by the native client. Each statistic has a type of either int, long,
or double, and either a gauge or a counter. The value of a gauge can increase and decrease, and
the value of a counter strictly increases. Create an instance of StatisticDescriptor by calling
one of these StatisticsFactory functions: createDoubleCounter, createDoubleGauge,
createIntCounter, createIntGauge ], createLongCounter, createLongGauge.
StatisticsType. An instance of this class describes a logical collection of StatisticDescriptors.
These descriptions are used to create an instance of Statistics. Create an instance of
StatisticsType by calling StatisticsFactory::createType.
Statistics. An instance of this class represents concrete Statistics of the associated
StatisticsType. This class stores data related to all individual statistic objects. Create an instance
by calling StatisticsFactory::createStatistics. This class has functions to get, set, and
increment statistic values.
StatisticsFactory. This class provides functions for creating instances of StatisticDescriptor ,
StatisticsType, and Statistics objects. This is a singleton class, and you acquire its instance
by using StatisticsFactory::getExistingInstance.
To create new statistics, see Creating New Statistics.
Creating a Cache
The code snippets in this section show cache creation.
When you create your cache, the system automatically connects your process to the server tier. For
systems with security enabled, the credentials for a connecting client are authenticated when it creates the
cache. See Security on page 158 for more information about authenticated connections.
88
Creating a Proxy Client-Side Region
Pivotal GemFire Native Client
Creating the System Connection and the Cache
In this example, the application creates the cache by calling the
CacheFactory::create function, specifying the servers to connect to:
CacheFactoryPtr cacheFactory =
CacheFactory::createCacheFactory();
CachePtr cachePtr = cacheFactory
->addServer("localhost", 40404)
->addServer("localhost", 40405)
->setSubscriptionEnabled(true)
->create();
Creating a Proxy Client-Side Region
This section provides code examples for creating and customizing proxy client-side regions.
Note: Creating a region through the native client API only creates a proxy client-side region. A
corresponding region with the same name and path should also exist on the servers that have been
configured for client connections and upon which the client will perform its operations.
To create a region, you create a RegionFactory using the RegionShortcut that most closely fits
your region configuration. From that, create your region, customizing the settings as region attributes as
needed.
Creating a CACHING_PROXY Region
This example creates a region using a CACHING_PROXY RegionShortcut with no
further modifications:
RegionFactoryPtr regionFactory =
cachePtr->createRegionFactory(CACHING_PROXY);
regionPtr = regionFactory ->create("exampleRegion");
Creating a CACHING_PROXY Region with LRU
This example creates a region based on the CACHING_PROXY RegionShortcut with two
additional region attributes settings. For information on the settings, see Region Attributes
Descriptions.
RegionFactoryPtr regionFactory =
cachePtr->createRegionFactory(CACHING_PROXY);
regionPtr = regionFactory->setLruEntriesLimit( 20000 )
->setInitialCapacity( 20000 )
->create("exampleRegion");
89
Adding an Entry to the Cache
Pivotal GemFire Native Client
Adding an Entry to the Cache
You can populate a native client region with cache entries using the Region::put or the
Region::create API functions. Code examples demonstrate these actions.
The put function places a new value into a region entry with the specified key, while the create function
creates a new entry in the region. Both functions provide a user-defined parameter object to any cache
writer invoked in the process, and new values for both functions are propagated to a connected cache
server.
Adding Entries Using create
When the put function adds an entry, the previous value is overwritten if there is already
an entry associated with the specified key in the region. In the next example, the program
uses the API to put 100 entries into the cache by iteratively creating keys and values, both
of which are integers.
for ( int32_t i=0; i < 100; i++ ) {
regionPtr->put( i, CacheableInt32::create(i) );
}
Bulk Put Operations Using putAll
You can batch up multiple key/value pairs into a hashmap and put them into the cache with a single
operation using the Region::putAll API function ( Region.PutAll for .NET) . Each entry is
processed for interest registration on the server, so each entry requires its own unique event ID. Updates
and creates can be mixed in a putAll operation, so those events need to be addressed on the cache
server for appropriate cache listener invocation on distributed system members. Map entries retain their
original order when they are processed at the server.
The following table lists the client and cache server statistics for putAll.
Table 14: PutAll Statistics for Cache Server and Client
Statistic Type
Chart Name
Description
CachePerfStats
Putalls
Total number of times a map is
added or replaced in the cache as
a result of a local operation. Also
reports the number of putAll
operations.
CacheperfStats
putallTime
Total time to replace a map in
the cache as a result of a local
operation.
CacheServerStats
putAllRequests
Number of putAll requests.
CacheServerStats
putAllResponses
Number of putAll responses
written to the cache client.
CacheServerStats
processPutAllTime
Total time to process a cache
client putAll request, including
the time to put all objects into the
cache.
90
Accessing an Entry
Pivotal GemFire Native Client
Statistic Type
Chart Name
Description
CacheServerStats
readPutAllRequestTime
Total time to read putAll
requests.
CacheServerStats
writePutAllResponseTime
Total time to write putAll
responses.
CacheClientStats
putAll
Number of putAll requests sent
to the cache server.
CacheClientStats
sendPutAllTime
Total time for sendPutAll .
Accessing an Entry
The standard Region::get API method returns the value associated with the specified key, and passes
the callback argument to any cache loaders or cache writers that are invoked in the operation.
If the value is not present locally, it is requested from the cache server. If the cache server request is
unsuccessful, a local cache loader is invoked.
The entry value is either retrieved from the local cache or fetched by the region’s locally defined cache
loader.
In the following example, the program uses the API to do a get for each entry that was put into the cache:
for ( int32_t i=0; i< 100; i++) {
CacheableInt32Ptr res = dynCast<CacheableInt32Ptr>(regionPtr->get(i));
}
Bulk Get Operations Using getAll
You can use the Region::getAll API ( Region.GetAll for .NET) to gets values for an array of keys
from the local cache or server. If the value for a key is not present locally, then it is requested from the
server.
Note: The value returned is not copied, so multi-threaded applications should not modify the value
directly, but should instead use the update methods.
See the Region online API documentation for more information about using getAll.
Serializing Data
All data that Pivotal GemFire moves out of the local cache must be serializable.
Region Data Requiring Serialization
Certain region types (including client regions) require serialization.
Region data in the following types of regions must be serializable:
•
•
Partitioned regions (except functions that add data locally to a partitioned region use the deserialized
form).
Distributed regions.
91
Pivotal GemFire Native Client C++ API
•
•
•
•
•
Pivotal GemFire Native Client
Regions that are persisted or overflowed to disk.
Server or client regions in a client/server installation.
Regions distributed between gateways in a multi-site installation.
Regions that receive events from remote caches.
Regions that provide function arguments and results.
To minimize the cost of serialization and deserialization, GemFire avoids changing the data format
whenever possible. This means your data may be stored in the cache in serialized or deserialized form,
depending on how you use it. For example, if a server acts only as a storage location for data distribution
between clients, it makes sense to leave the data in serialized form, ready to be transmitted to clients that
request it. Partitioned region data is always stored in serialized form with one exception—functions that
add data to a partitioned region locally use the deserialized form.
Data Serialization Options
The native client C++ API gives you two serialization options: the GemFire::Serializable interface and
GemFire PDX serialization.
GemFire Portable Data eXchange (PDX) serialization is the recommended option. GemFire PDX
serialization provides portability for PDX serializable objects so that clients can share data with Java
servers and other non-C++ clients. PDX is a cross-language data format that can reduce the cost of
distributing and serializing your objects. PDX stores data in named fields that you can access individually
in order to avoid the cost of deserializing the entire data object. PDX also allows you to mix versions of
objects where you have added or removed fields.
When using PDX serialization, you can either use the PdxSerializer (for all your domain objects) or
PdxSerializable (for a specific domain object).
PdxSerializer is used when a user has registered his or her domain class for serialization in the cache
using the registerPdxSerializer API.
PdxSerializable is used when the domain class that a user wants to serialize/deserialize is
inherited from PdxSerializable interface, and the user has registered domain class using
registerPdxType(domainClass) API.
The non-PDX serialization option is to use the GemFire::Serializable interface. The
GemFire::Serializable interface can be a good option performance-wise if the size of your objects
is small. The GemFire::Serializable is used whenever a user domain class is not inherited by
PdxSerializable but the user has registered his or her class with the registerType API. (See
Serializing Data with the GemFire Serializable Interface on page 102 for more information.)
Table 15: Serialization Options: Comparison of Features
Capability
GemFire::Serializable
GemFire::PdxSerializable
Handles multiple versions of
domain objects*
X
Provides single field access
on servers of serialized data,
without full deserialization.
Supported also for OQL
queries.
X
Automatically ported to other
languages by GemFire - no
X
92
Pivotal GemFire Native Client C++ API
Capability
Pivotal GemFire Native Client
GemFire::Serializable
GemFire::PdxSerializable
X
X (See Using PDX Serialization with
Delta Propagation on page 102 for
requirements.)
need to program Java-side
implementation
Works with GemFire delta
propagation
*You can mix domain object versions where the differences between versions are the addition and removal
of object fields.
For detailed information on the interfaces, see the online API documentation.
Serializing Data with GemFire PDX Serialization
When you use GemFire PDX serialization with the native client C++ API, you can register a PdxSerializer
for the entire cache or implement PDX serialization for each domain object.
You can also set the object preference of the cache to the PdxInstance type, which allows you to access
fields of a PDX object without deserializing the entire object.
Note: Autoserialization (also referred to as the PDX autoserializer) is not currently available for the
native client C++ API.
Serialize Your Domain Objects with PdxSerializer and
PdxWrapper
For domain objects that you cannot or do not want to modify, use the PdxSerializer and the
PdxWrapper classes to serialize and deserialize the object's fields.
You register a PdxSerializer implementation for the entire cache, programming it for all of the domain
objects that you handle in this way. This way you do not have to implement the PdxSerializable
interface for each domain class.
The PdxSerializer allows domain classes to be serialized and deserialized as PDXs without
modification of the domain class. It only requires that the domain class have a constructor accessible to the
PdxSerializer to create an instance. The domain class will be held in a wrapper class PdxWrapper.
PdxSerializer has the following methods:
•
•
The toData method returns true if the PdxSerializer was able to serialize userObject, false if not.
If the PdxSerializer was able to deserialize the object, the fromData method returns a void pointer to
the user object to be wrapped in a PdxWrapper.
When you later reference the user object, use the PdxWrapper class. PdxWrapper holds a shared
reference to the object in the local cache and is used during serialization and deserialization. PdxWrapper
acts as a container for the user domain object and needs to wrap every instance of the object that
uses a registered PdxSerializer. The object instance will not be modified. In addition, when using
PdxWrapper, you will need to provide a function pointer to a "de-allocator" which will delete the user
object when the reference is no longer held.
The following code example defines a user object and a PdxSerializer. It then registers the new
PdxSerializer and then uses PdxWrapper to put the object in a region and retrieve the object from a
region.
class UserClass
{
public:
93
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
int m_int;
string m_string;
UserClass(int intVal, string stringVal)
{
m_int = intVal;
m_string = stringVal;
}
static void deallocate(void * object, char * className)
{
if (strcmp(className, "com.example.UserClass") == 0) {
UserClass * userObject = reinterpret_cast<UserClass*>(object);
delete userObject;
}
}
};
class UserPdxSerializer : public PdxSerializer
{
public:
void* fromData(char * className, PdxReaderPtr pdxReader)
{
if (strcmp(className, "com.example.UserClass") != 0) {
return NULL;
}
int intVal = pdxReader->readInt("m_int");
string stringVal = pdxReader->readString("m_string");
UserClass * userObject = new UserClass(intVal, stringVal);
}
return (void*) userObject;
bool toData(void * object, char * className, PdxWriterPtr pdxWriter)
{
if (strcmp(className, "com.example.UserClass") != 0) {
return false;
}
UserClass * userObject = reinterpret_cast<UserClass*>(object);
pdxWriter->writeInt("m_int", userObject->m_int);
pdxWriter->writeString("m_string", userObject->m_string);
}
return true;
UserDeallocator getDeallocator(char * className)
{
if (strcmp(className, "com.example.UserClass") == 0) {
return UserClass::deallocate;
} else {
return NULL;
}
}
};
// Register a user PDX serializer
Serializable::registerPdxSerializer(new UserPdxSerializer);
94
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
// Put a user object into a region.
UserClass * userObject = new UserClass(123, "someValue");
PdxWrapperPtr pdxWrapper = new PdxWrapper(userObject,
"com.example.UserClass",
UserClass::deallocate);
region->put("key", pdxWrapper);
// Get a user object from a region.
pdxWrapper = dynCast<PdxWrapperPtr>(region->get("key"));
UserClass * userObject = reinterpret_cast<UserClass*>(pdxWrapper>getObject());
Serialize Using the GemFire PdxSerializable Class
Domain classes need to inherit the PdxSerializable abstract class to serialize and de-serialize the
object. When you write objects using PDX serialization, they are distributed to the server tier in PDX
serialized form.
When you run queries against the objects on the servers, only the fields you specify are deserialized. A
domain class should serialize and de-serialize all its member fields in the same order in its toData and
fromData method.
Use this procedure to program your domain object for PDX serialization using the PdxSerializable
abstract class.
1. In your domain class, implement PdxSerializable. Example:
class PdxObject: public PdxSerializable
2. Program the toData function to serialize your object as required by your application.
If you also use PDX serialization in Java or .NET for the object, serialize the object in the same way for
each language. Serialize the same fields in the same order and mark the same identity fields.
3. Program the fromData method to read your data fields from the serialized form into the object's fields.
In your fromData implementation, use the same name as you did in toData and call the read
operations in the same order as you called the write operations in your toData implementation.
4. Optionally, program your domain object's hashCode and equals methods.
Use the markIdentityField method to indicate that the given field name should be included in
hashCode and equals checks of this object on a server.
The fields that are marked as identity fields are used to generate the hashCode and equals methods of
PdxInstance. Because of this, the identity fields should themselves either be primitives, or implement
hashCode and equals.
If no fields are set as identity fields, then all fields will be used in hashCode and equals checks. The
identity fields should make marked after they are written using a write* method.
PdxSerializable Example
class PdxObject: public PdxSerializable {
private:
uint32_t m_id;
char* m_str;
public:
PdxObject(){};
95
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
PdxObject(uint32_t id, char* str);
virtual ~PdxObject();
uint32_t getID() {
return m_id;
}
char* getStr(){
return m_str;
}
virtual void toData(PdxWriterPtr pw) const;
virtual void fromData(PdxReaderPtr pr);
CacheableStringPtr toString() const;
virtual char* getClassName() const;
static Cacheable* createDeserializable() {
return new PdxObject();
}
};
PdxObject::PdxObject(uint32_t i, char* str) {
m_id = i;
m_str = str;
}
PdxObject::~PdxObject() {
}
void PdxObject::toData( PdxWriterPtr pw ) const {
pw->writeInt("id", m_id);
pw->markIdentityField("id");
pw->writeString("str", m_str);
}
void PdxObject::fromData( PdxReaderPtr pr )
{
m_id = pr->readInt("id");
m_str = pr->readString("str");
}
char* getClassName() const{
{
return "com.example.PdxType";
}
CacheableStringPtr PdxObject::toString() const {
char idbuf[1024];
sprintf(idbuf,"PdxObject: [ ID=%d ]",m_id);
return CacheableString::create( idbuf );
}
Performing put, get, and localDestroy Operations with a PDX Domain
Object
This topic demonstrates how you can perform operations on a PDX domain object after you have
implemented PDX serializable in your domain class.
For example, you can perform operations like put, get, and localDestroy with the domain class you defined
for PDX serialization in the PdxSerializable Example on page 95.
To perform operations, you could write the following application code:
96
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
1. Register the PDX domain class.
Serializable::registerPdxType(PdxObject::createDeserializable);
2. Create the PDX domain object PdxObject.
CacheablePtr pdxobj(new PdxObject(100, "Value-1"));
CacheableKeyPtr keyport = CacheableKey::create("ABC");
3. Here's an example of a put operation.
rptr->put(keyport, pdxobj);
4. Here's an example of locally destroying the entry.
rptr->localDestroy(keyport);
5. Here's an example of a get operation.
PdxObject *obj2 = dynamic_cast<PdxObject *> ((rptr->get(keyport)).ptr());
LOGINFO("Debug:Returned ID = %d", obj2->getID());
Programming Your Application to Use PdxInstances
A PdxInstance is a lightweight wrapper around the raw bytes of the PDX serialized objects kept in the
cache. It provides applications with run-time access to files of a PDX serialized object. GemFire provides
the implementation of the PdxInstance class.
You can configure your cache to return a PdxInstance when a PDX serialized object is deserialized
instead of deserializing the object to a domain class. Preventing deserialization saves both time and
memory and does not require you deserialize the object to the domain class.
This configuration can be done in cache.xml by setting the attribute read-serialized
to true on the <pdx>element. Or it can be done programmatically using the
CacheFactory::setPdxReadSerialized(bool) method.
After this preference is configured, any time a PDX object is deserialized, it is deserialized into a
PdxInstance.
The following is a code sample of using the setField API of PdxInstance to modify fields:
RegionPtr rptr = getHelper()->getRegion( regionNames[0] );
CacheableKeyPtr keyport = CacheableKey::create("pdxput");
CacheableKeyPtr keyport1 = CacheableKey::create("pdxput2");
PdxInstancePtr pIPtr = dynCast<PdxInstancePtr>(rptr->get(keyport));
LOG( "modifyPdxInstance get complete." );
WritablePdxInstancePtr wpiPtr( pIPtr->createWriter());
ASSERT(pIPtr != NULLPTR, "pIPtr != NULLPTR expected");
int val = 0;
int newVal = 0;
ASSERT(pIPtr->hasField("m_int32") == true, "m_id1 = true expected");
pIPtr->getField("m_int32", val);
wpiPtr->setField("m_int32", val + 1);
rptr->put(keyport, wpiPtr);
PdxInstancePtr newPiPtr = dynCast<PdxInstancePtr>(rptr->get(keyport));
ASSERT(newPiPtr->hasField("m_int32") == true, "m_int32 = true expected");
newPiPtr->getField("m_int32", newVal);
ASSERT(val + 1 == newVal, "val + 1 == newVal expected");
ASSERT((*pIPtr.ptr() == *newPiPtr.ptr()) == false,
97
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
"PdxInstance should not be equal");
In addition to field access, PdxInstance also supports field modification using the
setField(fieldName) method. The setField method has copy-on-write semantics. So for the
modifications to be stored in the cache, the PdxInstance must be put into a region after setField has
been called one or more times.
Configuring PDX to Ignore Unread Fields During Deserialization
Use the setPdxIgnoreUnreadFields API to control whether PDX ignores fields that were unread
during deserialization.
Use the setPdxIgnoreUnreadFields API to control whether PDX ignores fields that were unread
during deserialization.
The default is to preserve unread fields by including their data during serialization. However, if you
configure the cache to ignore unread fields then their data will be lost during serialization.
You should only set this attribute to true if you know this member will only be reading cache data. In this
use case you do not need to pay the cost of preserving unread fields since you will never reserialize the
PDX data.
For example:
CacheFactoryPtr cfPtr = CacheFactory::createCacheFactory(PropertiesObj);
cfPtr->setPdxReadSerialized(tue);
cfPtr->setPdxIgnoreUnreadFields(false);
cachePtr = cfPtr->create();
Using PdxInstanceFactory to Create PdxInstances
You can use the PdxInstanceFactory API to create a PdxInstance from raw data when the domain
class in not available on the server.
Creating a PdxInstance can be particularly useful when you need an instance of a domain class for plugin code such as a function or a loader. If you have raw data for the domain object (the class name and
each field's type and data), then you can explicitly create a PdxInstance. The PdxInstanceFactory
API is very similar to the PdxWriter API except that after writing each field, you need to call the create
method which returns the created PdxInstance.
PdxInstance Example
The following is a code example of creating a PdxInstance.
class Person
{
private:
char* m_name;
int m_id;
int m_age;
public:
Person() { }
Person(char* name, int id, int age)
{
m_name = name;
m_id = id;
m_age = age;
}
98
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
char* getName() const
{
return m_name;
}
int getID()
{
return m_id;
}
int getAge()
{
return m_age;
}
};
int main(int argc, char ** argv)
{
try
{
// Create a GemFire Cache.
CacheFactoryPtr cacheFactory =
CacheFactory::createCacheFactory();
CachePtr cachePtr = cacheFactory->set("cache-xml-file",
"XMLs/clientPdxInstance.xml")->create();
LOGINFO("Created the GemFire Cache");
// Get the example Region from the Cache which is declared
in the
Cache XML file.
RegionPtr regionPtr = cachePtr->getRegion("Person");
LOGINFO("Obtained the Region from the Cache.");
Person* p = new Person("Jack", 7, 21);
//PdxInstanceFactory for Person class
PdxInstanceFactoryPtr pif = cachePtr>createPdxInstanceFactory("Person");
LOGINFO("Created PdxInstanceFactory for Person class");
pif->writeString("m_name", p->getName());
pif->writeInt("m_id", p->getID());
pif->markIdentityField("m_id");
pif->writeInt("m_age", p->getAge());
PdxInstancePtr pdxInstance = pif->create();
LOGINFO("Created PdxInstance for Person class");
regionPtr->put("Key1", pdxInstance);
LOGINFO("Populated PdxInstance Object");
PdxInstancePtr retPdxInstance = regionPtr->get("Key1");
LOGINFO("Got PdxInstance Object");
int id = 0;
retPdxInstance->getField("m_id", id);
int age = 0;
retPdxInstance->getField("m_age", age);
99
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
char* name = NULL;
retPdxInstance->getField("m_name", &name);
if(id == p->getID()&& age == p->getAge() && strcmp(name, p>getName()) == 0
&& retPdxInstance->isIdentityField("m_id") == true)
LOGINFO("PdxInstance returns all fields value expected");
else
LOGINFO("PdxInstance doesn't returns all fields value
expected");
delete p;
// Close the GemFire Cache.
cachePtr->close();
LOGINFO("Closed the GemFire Cache");
}
// An exception should not occur
catch(const Exception & gemfireExcp)
{
LOGERROR("PdxInstance GemFire Exception: %s",
gemfireExcp.getMessage());
}
}
Using C++ Enum Type with PDX Serialization
Because there is no "object" base type in C++, enums cannot be directly passed as a parameter to the
writeObject and readObject API.
To use the C++ enum type with PDX serialization, you will have to wrap the enum in the CacheableEnum
class type by specifying classname, enumname and ordinal.
enum enumQuerytest { id1, id2, id3 };
class TESTOBJECT_EXPORT PdxEnumTestClass :public PdxSerializable
{
private:
int m_id;
CacheableEnumPtr m_enumid;
public:
int getID(){
return m_id;
}
CacheableEnumPtr getEnumID() {
return m_enumid;
}
PdxEnumTestClass(int id)
{
m_id = id;
switch (m_id)
{
case 0:
m_enumid = CacheableEnum::create("enumQuerytest", "id1", id1);
break;
case 1:
100
Pivotal GemFire Native Client C++ API
}
}
Pivotal GemFire Native Client
m_enumid = CacheableEnum::create("enumQuerytest", "id2", id2);
break;
case 2:
m_enumid = CacheableEnum::create("enumQuerytest", "id3", id3);
break;
default:
m_enumid = CacheableEnum::create("enumQuerytest", "id1", id1);
break;
PdxEnumTestClass() { }
void toData(PdxWriterPtr pw) {
pw->writeInt("m_id", m_id);
pw->writeObject("m_enumid", m_enumid);
}
void fromData(PdxReaderPtr pr) {
m_id = pr->readInt("m_id");
m_enumid = pr->readObject("m_enumid");
}
CacheableStringPtr toString() const {
return CacheableString::create("PdxEnumTestClass");
}
char* GetClassName() const {
return "com.example.PdxEnumTestClass";
}
static PdxSerializable* createDeserializable() {
return new PdxEnumTestClass();
}
};
How Put and Queries Work on Enum
The following code sample demonstrates how put and query operations work when using the C++ enum
Type with PDX serialization:
//Creating objects of
PdxEnumTestClassPtr
PdxEnumTestClassPtr
PdxEnumTestClassPtr
type PdxEnumTestClass
pdxobj1(new PdxEnumTestClass(0));
pdxobj2(new PdxEnumTestClass(1));
pdxobj3(new PdxEnumTestClass(2));
RegionPtr rptr = getHelper()->getRegion( "DistRegionAck" );
//PUT Operations
rptr->put( CacheableInt32::create(0), pdxobj1 );
LOG( "pdxPut 1 completed " );
rptr->put( CacheableInt32::create(1), pdxobj2 );
LOG( "pdxPut 2 completed " );
rptr->put( CacheableInt32::create(2), pdxobj3 );
LOG( "pdxPut 3 completed " );
//Query
try {
101
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
Serializable::registerPdxType(PdxEnumTestClass::createDeserializable);
LOG("PdxEnumTestClass Registered Successfully....");
} catch (gemfire::IllegalStateException&/* ex*/) {
LOG("PdxEnumTestClass IllegalStateException");
}
RegionPtr rptr = getHelper()->getRegion( "DistRegionAck" );
SelectResultsPtr results = rptr->query("m_enumid.name = 'id2'");
ASSERT(results->size()== 1 , "query result should have one item");
ResultSetPtr rsptr = dynCast<ResultSetPtr>(results);
SelectResultsIterator iter = rsptr->getIterator();
while (iter.moveNext())
{
PdxEnumTestClassPtr re = dynCast<PdxEnumTestClassPtr>(iter.current());
ASSERT(re->getID()== 1 , "query should have return id 1");
Using PDX Serialization with Delta Propagation
To use delta propagation with PDX serialization, you must implement the Delta interface methods.
Users can include delta propagation support with PDX serialization by implementing the Delta interface
methods. However, using delta propagation with PDX will require the users to implement Java side classes
and the objects will remain in deserialized form at all times on the server and you will lose one of the main
benefits of PDX.
The following code snippet is a sample implementation of the Delta interface methods for using with PDX
serialization.
class PdxWithDelta : public PdxSerializable, public Delta
{
public:
bool hasDelta();
void toDelta(DataOutput& output);
void fromDelta(DataInput& input);
DeltaPtr clone();
// other PdxSerializable methods here...
};
Serializing Data with the GemFire Serializable Interface
The native client C++ API provides a Serializable interface that you can use for fast and compact
data serialization. This section discusses the GemFire serializable interface, and presents implementation
examples.
•
•
•
•
•
•
•
•
How Serialization Works on page 103
Implementing the Serializable Interface on page 103
Example 1: The Simple Class BankAccount on page 103
Example 2: Implementing a Serializable Class on page 104
Registering the Type on page 105
Custom Key Types on page 105
Example 3: Extending a Serializable Class To Be a CacheableKey on page 105
Serialization in Native Client Mode with a Java Server on page 106
102
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
How Serialization Works
When your application puts an object into the cache for subsequent distribution, GemFire serializes the
data by taking these steps:
1. Calls the appropriate classId function.
2. Writes the full typeId using the classId for the instance.
3. Invokes the instance’s toData function.
When your application subsequently receives a byte array, GemFire takes the following steps:
1. Decodes the typeId, extracts the classId from the typeId, then creates an object of the designated
type using the registered factory functions.
2. Invokes the fromData function with input from the data stream.
3. Decodes the data, then populates the data fields.
Implementing the Serializable Interface
To store your own data types in the cache, you need to derive a new subclass from the Serializable
interface. In practical terms, this means that you need to implement a small set of helper functions:
1. Write a toData function that serializes your data.
void toData (DataOutput& output)
The toData function is responsible for copying all of the object’s data fields to the object stream.
The DataOutput class represents the output stream and provides methods for writing the primitives in
a network byte order. For more about this, see the online API documentation for DataOutput .
The toData function is responsible for copying all of the object’s data fields to the object stream.
The DataOutput class represents the output stream and provides methods for writing the primitives in
a network byte order. For more about this, see the online API documentation for DataOutput .
2. Write a fromData function that consumes a data input stream and repopulates the object’s data fields.
void fromData (DataInput& input)
The DataInput class represents the input stream and provides methods for reading input elements.
The fromData function must read the elements of the input stream in the same order that they were
written by toData . For more about this, see the online API documentation for DataInput .
Example 1: The Simple Class BankAccount
This example demonstrates a simple BankAccount class that encapsulates two ints :
ownerId and accountId :
class BankAccount
{
private:
int m_ownerId;
int m_accountId;
public:
BankAccount( int owner, int account )
: m_ownerId( owner ),
m_accountId( account )
{
}
103
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
int getOwner( )
{
return m_ownerId;
}
int getAccount( )
{
return m_accountId;
}
};
To make BankAccount serializable, you would need to derive the class from
Serializable and implement the following:
•
•
•
•
toData—a function to serialize the data.
fromData—a function to deserialize the data.
classId—a function to provide a unique integer for the class.
TypeFactoryMethod—a pointer to a function that returns a Serializable* to an
uninitialized instance of the type.
Example 2: Implementing a Serializable Class
This example shows a code sample that demonstrates how to implement a serializable
class.
class BankAccount
: public Serializable
{
private:
int m_ownerId;
int m_accountId;
public:
BankAccount( int owner, int account )
: m_ownerId( owner ),
m_accountId( account )
{
}
int getOwner( )
{
return m_ownerId;
}
int getAccount( )
{
return m_accountId;
}
// Add the following for the Serializable interface
// Our TypeFactoryMethod
static Serializable* createInstance( )
{
return new BankAccount( 0, 0 );
}
int32_t classId( )
{
return 10; // must be unique per class.
}
virtual uint32_t objectSize() const
{
return 10;
}
void toData( DataOutput& output )
104
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
{
output.writeInt( m_ownerId );
output.writeInt( m_accountId );
}
Serializable* fromData( DataInput& input )
{
input.readInt( &m_ownerId );
input.readInt( &m_accountId );
return this;
}
};
Registering the Type
To be able to use the BankAccount type, you must register it with the type system so that when
an incoming stream contains a BankAccount , it can be manufactured from the associated
TypeFactoryMethod .
Serializable::registerType( BankAccount::createInstance );
Typically, you would register the type before calling the function DistributedSystem::connect.
Note: Type IDs must be unique to only one class.
Custom Key Types
If your application uses key types that are too complex to easily force into CacheableString, you can
likely improve performance by deriving a new class from CacheableKey. If you have hybrid data types
you can implement your own derivation of CacheableKey that encapsulates the data type.
See Serialization in Native Client Mode with a Java Server for information about implementing key types
for a native client that is used with a Java cache server.
To extend a Serializable class to be a CacheableKey, you need to modify the class definition as
follows:
•
•
Change the class so that it derives from CacheableKey rather than Serializable.
Implement operator== and hashcode functions.
The next example demonstrates how to extend a serializable class to be a cacheable key.
Example 3: Extending a Serializable Class To Be a CacheableKey
This example shows how to extend a serializable class to be a cacheable key.
class BankAccount
: public CacheableKey
{
private:
int m_ownerId;
int m_accountId;
public:
BankAccount( int owner, int account )
: m_ownerId( owner ),
m_accountId( account )
{
}
int getOwner( )
{
return m_ownerId;
105
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
}
int getAccount( )
{
return m_accountId;
}
// Our TypeFactoryMethod
static Serializable* createInstance( )
{
return new BankAccount( 0, 0 );
}
int32_t typeId( )
{
return 1000; // must be unique per class.
}
void toData( DataOutput& output )
{
output.writeInt( m_ownerId );
output.writeInt( m_accountId );
}
Serializable* fromData( DataInput& input )
{
input.readInt( &m_ownerId );
input.readInt( &m_accountId );
return this;
}
// Add the following for the CacheableKey interface
bool operator == ( const CacheableKey& other ) const
{
const BankAccount& otherBA =
static_cast<const BankAccount&>( other );
return (m_ownerId == otherBA.m_ownerId) &&
(m_accountId == otherBA.m_accountId);
}
uint32_t hashcode( ) const
{
return m_ownerId;
}
virtual int32_t classId( )const
{
return 10; // must be unique per class.
}
virtual uint32_t objectSize() const
{
return 10;
}
};
Serialization in Native Client Mode with a Java Server
Primitive object types supported in all languages (CacheableInt32, CacheableString,
CacheableBytes) function without requiring custom definitions with the Java cache server. For the keys,
the Java cache server has to deserialize them and locate the hashcode to be able to insert the internal
maps. Because of this, key types for C++ and .NET native clients used with a Java server are required to
be registered on the Java server, but the value types do not need to be registered. This needs to be done
even if there are no Java clients. The Java serializable type should have the same classId as the .NET
class, and it should serialize and deserialize the type in the same manner as the .NET implementation.
See Data Serialization for more information about .NET data serialization.
106
Implementing User-Defined Objects in Java Clients
Pivotal GemFire Native Client
Serializing Object Graphs
If you have a graph of objects where each node can be serializable, the parent node can call
DataOutput::writeObject to delegate the serialization responsibility to its child nodes. Similarly, your
application can call DataInput::readObject to deserialize the object graph.
For more information, see the online API documentation for DataOutput and DataInput.
Serializing and Accessing Data as a Blob
If you have data that is best handled as a blob, such as structs that do not contain pointers, use the
serializable type CacheableBytes . CacheableBytes is a blob class that implements the serialization
for you.
CacheableBytes also provides direct access to the blob data. Because it is not derived from the
CacheableKey interface, CacheableBytes enables you to modify data in place and then put it into the
region again to distribute the change.
For more information, see the online API documentation for CacheableBytes.
Implementing User-Defined Objects in Java Clients
You can use one of two methods to implement a user-defined object in a Java client that works with C++
clients: Instantiator.register and DataSerializable.
Instantiator.register
With the Instantiator.register method, a client sends a RegistrationMessage to every Java VM
in its distributed system. The message announces the mapping between a user-defined classId and class
name. The other JVMs can deserialize the byte array with the correct class.
If two clients are in different distributed systems, a RegistrationMessage cannot be sent between
them. For example: a put made by a client in one distributed system will hang when a client in another
distributed system performs a get in pure Java mode. Similarly, a put made by a C++ client will cause a
Java client to hang.
DataSerializable
Using the DataSerializable method, the user-defined object is serialized into the following byte array:
45 <2-byte-length> <class-name>
A Java client can deserialize the byte array, but a C++ client cannot convert the Java class name to a C++
class name.
Implementation
The DataSerializable method does not support using a nested object, while
Instantiator.register does support the use of nested objects. A workaround is to let each Java
client manually initiate an object for each possible user object class a C++ client provides, using the
following code:
User u = new User("", 0);
107
Using a Custom Class
Pivotal GemFire Native Client
See Java Serialization Example for a code sample that shows how to set up user object classes in a Java
client.
Using a Custom Class
This example shows how to use the defined BankAccount custom key type and the AccountHistory
value type.
The example takes you through these basic operations: registering, creating a cache, connecting to the
distributed system, putting data, getting data, and closing the cache.
Using a BankAccount Object
#include <gfcpp/GemfireCppCache.hpp>
#include "BankAccount.hpp"
#include "AccountHistory.hpp"
using namespace gemfire;
/*
This example connects, registers types, creates the cache,
creates a
region, and then puts and gets user defined type BankAccount.
*/
int main( int argc, char** argv )
{
// Register the user-defined serializable type.
Serializable::registerType( AccountHistory::createDeserializable );
Serializable::registerType( BankAccount::createDeserializable );
CacheFactoryPtr cacheFactory =
CacheFactory::createCacheFactory();
// Create a cache.
CachePtr cachePtr = cacheFactory->setSubscriptionEnabled(true)
->addServer("localhost", 24680)
->create();
// Create a region.
RegionFactoryPtr regionFactory =
cachePtr->createRegionFactory(CACHING_PROXY);
RegionPtr regionPtr = regionFactory->create("BankAccounts");
// Place some instances of BankAccount cache region.
BankAccountPtr KeyPtr(new BankAccount(2309, 123091));
AccountHistoryPtr ValPtr(new AccountHistory());
ValPtr->addLog( "Created account" );
regionPtr->put( KeyPtr, ValPtr );
printf( "Put an AccountHistory in cache keyed with BankAccount.
\n" );
// Call custom behavior on instance of BankAccount.
KeyPtr->showAccountIdentifier();
// Call custom behavior on instance of AccountHistory.
ValPtr->showAccountHistory();
// Get a value out of the region.
AccountHistoryPtr historyPtr =
dynCast<AccountHistoryPtr>( regionPtr->get( KeyPtr ) );
if ( historyPtr != NULLPTR ) {
printf( "Found AccountHistory in the cache.\n" );
historyPtr->showAccountHistory();
historyPtr->addLog( "debit $1,000,000." );
regionPtr->put( KeyPtr, historyPtr );
printf( "Updated AccountHistory in the cache.\n" );
108
Creating New Statistics
Pivotal GemFire Native Client
}
// Look up the history again.
historyPtr = dynCast<AccountHistoryPtr>( regionPtr>get( KeyPtr ) );
if ( historyPtr != NULLPTR ) {
printf( "Found AccountHistory in the cache.\n" );
historyPtr->showAccountHistory();
}
// Close the cache and disconnect from the servers
cachePtr->close();
return 0;
}
Creating New Statistics
This example provides a programmatic code sample for creating and registering new statistics.
For information about the gemfire_statistics API, see Statistics API.
Creating New Statistics Programmatically
//Get StatisticsFactory
StatisticsFactory* factory =
StatisticsFactory::getExistingInstance();
//Define each StatisticDescriptor and put each in an array
StatisticDescriptor** statDescriptorArr = new
StatisticDescriptor*[6];
statDescriptorArr[0] = statFactory>createIntCounter("IntCounter",
"Test Statistic Descriptor Int Counter.","TestUnit");
statDescriptorArr[1] = statFactory->createIntGauge("IntGauge",
"Test Statistic Descriptor Int Gauge.","TestUnit");
statDescriptorArr[2] = statFactory>createLongCounter("LongCounter",
"Test Statistic Descriptor Long Counter.","TestUnit");
statDescriptorArr[3] = statFactory->createLongGauge("LongGauge",
"Test Statistic Descriptor Long Gauge.","TestUnit");
statDescriptorArr[4] = statFactory>createDoubleCounter("DoubleCounter",
"Test Statistic Descriptor Double Counter.","TestUnit");
statDescriptorArr[5] = statFactory>createDoubleGauge("DoubleGauge",
"Test Statistic Descriptor Double Gauge.","TestUnit");
//Create a StatisticsType
StatisticsType* statsType = statFactory>createType("TestStatsType",
"Statistics for Unit Test.",statDescriptorArr, 6);
//Create Statistics of a given type
Statistics* testStat =
109
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
factory->createStatistics(statsType,"TestStatistics");
//Statistics are created and registered. Set and increment
individual values
Int statIdIntCounter = statsType->nameToId("IntCounter");
testStat->setInt(statIdIntCounter, 10 );
testStat->incInt(statIdIntCounter, 1 );
int currentValue = testStat->getInt(statIdIntCounter);
110
Pivotal GemFire Native Client C++ API
Pivotal GemFire Native Client
111
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
Part
VII
Pivotal GemFire Native Client C# .NET API
Topics:
•
•
•
•
•
•
•
•
•
•
•
About the Native Client C# .NET
API
C++ Class to .NET Class
Mappings
Java to .NET Type Mapping
Table
Object Lifetimes
.NET Application Domains
Creating a Cache
Creating a Region
Adding an Entry to the Cache
Accessing an Entry
Data Serialization
Troubleshooting .NET
Applications
Pivotal GemFire Native Client C# .NET API describes the primary
classes, usage conventions, and C++ to .NET class mappings of the
native client .NET API. It demonstrates how to use the API to create
caches and perform data serialization.
112
About the Native Client C# .NET API
Pivotal GemFire Native Client
About the Native Client C# .NET API
The Microsoft .NET Framework interface for the Pivotal GemFire native client provides complete access
to the native client C++ functionality from any .NET Framework language (C#, C++/CLI, VB.NET, and J#).
This enables clients using C# and other .NET languages to use the capabilities provided by the GemFire C
++ API.
The GemFire native client uses a set of assemblies managed by the C++ Common Language
Infrastructure (C++ CLI). C++ CLI includes the libraries and objects necessary for common language types,
and it is the framework for .NET applications.
The .NET API for the native client adds .NET Framework CLI language binding for the GemFire native
client product.
Using C#, you can write callbacks and define user objects in the cache. The following figure shows an
overview of how a C# application accesses the native client C++ API functionality through C++/CLI .
Figure 5: C# .NET Application Accessing the C++ API
Note: This chapter uses C# as the reference language, but other .NET languages work the same
way.
The GemFire C# .NET API is provided in the GemStone::GemFire::Cache::Generic namespace. This
namespace allows you to manage your cache, regions, and data using the .NET Generics APIs.
Use the GemFire C# .NET API to programmatically create, populate, and manage a GemFire distributed
system.
Note: The C# .NET library is thread-safe except where otherwise indicated in the online API
documentation.
For complete and current information on the GemFire APIs, see the online .NET API documentation. For
general information on .NET, see the Microsoft developer’s network website.
113
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
C# .NET Naming and Usage Conventions
Unless noted, the .NET API classes and functions have the same names as their C++ counterparts in the
namespace GemStone::GemFire::Cache. In .NET, all method names start with a capital letter.
The .NET interface names match those of comparable C++ interfaces, but with an ’I’ prepended to
satisfy .NET naming conventions. For example, the .NET equivalent of the C++ CacheLoader interface is
ICacheLoader.
The name of the GemFire Serializable interface is IGFSerializable because ISerializable is
a .NET built-in type.
Where possible, get* and set* functions are replaced by .NET properties.
You can implement the GemFire .NET interfaces. You cannot extend any of the classes because they are
marked as sealed.
Primary APIs in GemStone::GemFire::Cache::Generic
These are the main APIs used for cache, region, and data entry management in GemFire C# .NET. For
detailed information about the APIs, see the .NET API documentation included in the product or available
online.
Note: Declarative configuration via XML of application plug-ins such as cache listener, cache
writer, cache loader and partition resolver is not supported when clients are operated in the
new .NET Generic API.
Cache APIs
This section describes the CacheFactory and Cache classes.
•
•
CacheFactory class. Creates a Cache instance based on the provided distributed system and cache
configuration. Any gfcpp.properties and cache.xml files provided to the application are used to
initialize the system and cache. See Setting System and Cache Properties. If a cache.xml file is used
to create a cache and some of the regions already exist, a warning states that the regions exist and the
cache is created.
Cache class. Entry point to the GemFire caching API. This class allows you to create regions. The
cache is created by calling the create function of the CacheFactory class. When creating a cache,
you specify a DistributedSystem that tells the new cache where to find other caches on the network
and how to communicate with them.
Region and Entry APIs
This section describes classes for working with GemFire regions and region entries.
•
•
RegionFactory class. Creates a Region instance based on the provided configuration.
IRegion class. Provides functions for managing regions and cached data. The Region interface
implements the generic .NET IDictionary interface. In the GemFire APIs, IRegion implements
IDictionary and Region inherits IRegion, giving you access to the full range of .NET Generic
collection functions. Use the functions in this class to perform the following actions:
•
•
•
•
Retrieve information about the region, such as its parent region and region attribute objects.
Invalidate or destroy the region.
Add, update, invalidate, and remove region entries.
Determine, individually or as entire sets, the region's entry keys, entry values, and RegionEntry
objects.
114
Pivotal GemFire Native Client C# .NET API
•
•
•
•
Pivotal GemFire Native Client
RegionEntry class. Contains the key and value for the entry, and provides all non-distributed entry
operations. The operations of this object are not distributed and do not affect statistics.
RegionShortcut class. Holds enum definitions for the most common region configurations. Start your
region configuration with a shortcut setting in the region attribute, refid. Then customize further using
the RegionAttributes.
RegionAttributes class. Holds all attribute values for a region and provides functions for retrieving all
attribute settings. This class can only be modified by the AttributesFactory class before region
creation, and the AttributesMutator class after region creation.
AttributesMutator class. Allows modification of an existing region's attributes for application plug-ins
and expiration actions. Each region has an AttributesMutator instance.
Data Serialization APIs
Use either IPdxSerializable or IGFSerializable for each region. Do not mix the two.
For more information on these options, see Data Serialization.
•
•
•
•
•
•
•
•
•
•
IPdxSerializable interface. Provides a flexible way to serialize your domain objects for cache storage
and transfer to the servers. This is a GemFire built-in serialization framework. See Data Serialization.
IPdxReader. Supplies operations for reading data from GemFire IPDXSerializable types.
IPdxWriter. Provides operations for writing data into GemFire IPDXSerializable types.
IPdxInstance. Instance of a PDX serialized object that you can use to access the object’s data without
having to deserialize the object first.
IPdxInstanceFactory. Allows you to build an IPdxInstance using raw data.
IPdxTypeMapper interface. Allows you to map .NET type names to Java type names when using PDX
serialization.
IGFSerializable interface. Superclass of one set of user objects that can be serialized and stored in
the cache. These are GemFire built-in serializable types. See Data Serialization.
Serializable class. Wraps the native C++ gemfire::Serializable objects as managed
IGFSerializable objects. Whenever native C++ clients and .NET clients interoperate and are part of
the same distributed system, the user-defined types that are put by the native C++ clients that have not
been defined in .NET are returned as objects of this class.
The API contains overloads for most Region methods and other methods that take Serializable
as a value and that are more optimized than the more generic IGFSerializable overloads. The
application prefers using these overloads whenever the base class of an object is Serializable.
DataInput. Supplies operations for reading primitive data values and user-defined objects from a byte
stream.
DataOutput. Provides operations for writing primitive data values and user-defined objects
implementing IGFSerializable to an integer.
Event Handling APIs
Code your event handlers to do minimal work before returning control to GemFire.
For example, a listener implementation may hand off the event to a thread pool that processes the event
on its thread rather than the listener thread. Exceptions thrown by the listeners are caught by GemFire and
logged.
•
•
•
RegionEvent class. Provides information about the event, such as what region the event originated in,
whether the event originated in a cache remote to the event handler, and whether the event resulted
from a distributed operation.
EntryEvent class. Provides all available information for the RegionEvent . It also provides entryspecific information, such as the old and new entry values and whether the event resulted from a load
operation.
ICacheLoader application callback interface. Loads data into a region.
115
C++ Class to .NET Class Mappings
•
Pivotal GemFire Native Client
•
ICacheWriter application callback interface. Synchronously handles region and entry events before
the events occur. Entry events are create, update, invalidate, and destroy. Region events are
invalidate and destroy. This class has the ability to abort events.
ICacheListener application callback interface. Asynchronously handles region and entry events.
Listeners receive notifications when entries in a region change, or when changes occur to the
region attributes themselves. Entry events are create, update, invalidate, and destroy.
Region events are invalidate and destroy. Multiple events can cause concurrent invocation
of ICacheListener methods. If event A occurs before event B, there is no guarantee that their
corresponding ICacheListener method invocations will occur in the same order.
Property Collections and Logging APIs
This section describes classes for property collections and logging.
•
•
Properties class. Provides a collection of properties, each of which is a key/value pair. Each key is a
string, and the value can be a string or an integer.
Log class. Defines methods available to clients that need to write a log message to their GemFire
system shared log file. Any attempt to use an instance after its connection is disconnected throws a
NotConnectedException. For any logged message the log file contains:
•
•
•
•
The log level of the message.
The time the message was logged.
The ID of the connection and thread that logged the message.
The message itself, possibly with an exception including its stack trace.
C++ Class to .NET Class Mappings
Wherever the native C++ class methods use pass-by-reference semantics to return data, the
corresponding .NET methods return the object instead of using pass-by-reference semantics.
Table 16: C++ Class to .NET Class Mappings
C++ Class
.NET Class
class gemfire::AttributesFactory
Sealed class AttributesFactory
class gemfire::AttributesMutator
Sealed class AttributesMutator
class gemfire::Cache
Sealed class Cache
abstract class gemfire:: Cacheable
Interface IPdxSerializable or interface
IGFSerializable
class gemfire::CacheableBytes
Byte[] or ArrayList<Byte>
class gemfire::Cacheableint32
Int32
class gemfire:: CacheableString
String
abstract class gemfire:: CacheableKey
You can use any type that implements hashcode
and equals. The generic .NET built-in types are all
suitable.
abstract class gemfire::CacheListener
Interface ICacheListener
116
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
C++ Class
.NET Class
class gemfire:: CacheLoader
Interface ICacheLoader plus static class
CacheLoader
class gemfire:: CacheWriter
Interface class ICacheWriter
class gemfire::CacheFactory
Sealed class CacheFactory
class gemfire::DataInput
With IPdxSerializable, IPdxReader.
With IGFSerializable, sealed class
DataInput.
class gemfire:: DataOutput
With IPdxSerializable, IPdxWriter.
With IGFSerializable, sealed class
DataOutput .
class gemfire:: DiskPolicyType
enum DiskPolicyType plus static class
DiskPolicy containing convenience methods for
DiskPolicyType enumeration
class gemfire:: DistributedSystem
Sealed class DistributedSystem
class gemfire:: EntryEvent
Sealed class EntryEvent
class gemfire:: Exception
Class GemfireException
all other exceptions deriving from gemfire::
Exception
Corresponding exceptions deriving from
GemfireException
class gemfire:: ExpirationAction
enum ExpirationAction plus static class
Expiration containing convenience methods for
ExpirationAction enumeration
class gemfire:: Log
Static class Log. The native Log::log method is
mapped to Log.Write to avoid the conflict with the
class name which is reserved for the constructors
of Log class. The various loglevel Throw or Catch
methods are not implemented, since they are
redundant to Log::Log , Log::LogThrow, and
Log::LogCatch methods that take LogLevel as
a parameter.
enum gemfire:: MemberType
enum MemberType
abstract class gemfire::
PersistanceManager
Not provided. You can register
a C++ implementation using
AttributesFactory.SetPersistenceManager
but you cannot implement a new one in .NET
class gemfire:: Properties
Sealed class Properties
class gemfire:: Properties::Visitor
Delegate PropertiesVisitor
117
Java to .NET Type Mapping Table
Pivotal GemFire Native Client
C++ Class
.NET Class
abstract class gemfire:: Region
Class IRegion
class gemfire:: RegionAttributes
Sealed class RegionAttributes
class gemfire:: ScopeType
enum ScopeType plus static class Scope
containing convenience methods for ScopeType
enumeration+
abstract class gemfire:: Serializable
Two options:
•
•
Interface IPdxSerializable
Interface IGFSerializable plus
wrapper Serializable class for native
Serializable and UserData objects. The
native toString method is not provided, since
the ToString method of the base object class
provides the same functionality.
class gemfire:: SystemProperties
Sealed class SystemProperties
class gemfire:: UserData
Two options:
•
•
class gemfire:: VectorT<T>
Interface IPdxSerializable
Interface IGFSerializable
Array of the given type, such as T[]
Java to .NET Type Mapping Table
The following table provides a mapping between Java and .NET types.
Table 17: Java types and .NET types
Java Type
.NET Type
instances of PdxSerializable
.NET class of same name
instances of PdxInstance
.NET class of same name
instances serialized by a PdxSerializer
.NET class of same name
java.lang.Byte
System.SByte
java.lang.Boolean
System.Boolean
java.lang.Character
System.Char
java.lang.Short
System.Int16
java.lang.Integer
System.Int32
118
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
Java Type
.NET Type
java.lang.Long
System.Int64
java.lang.Float
System.Float
java.lang.Double
System.Double
java.lang.String
System.String
java.util.Date
System.DateTime
byte[]
System.Byte[]
boolean[]
System.Boolean[]
char[]
System.Char[]
short[]
System.Int16[]
int[]
System.Int32[]
long[]
System.Int64[]
float[]
System.Float[]
double[]
System.Double[]
String[]
System.String[]
byte[][]
System.Byte[][]
Object[]
system.Collections.Generic.List<Object>
java.util.HashMap
System.Collections.Generics.IDictionary<Object,
Object>
java.util.Hashtable
System.Collections.Hashtable
java.util.ArrayList
System.Collections.Generic.IList<Object>
java.util.Vector
Collections.ArrayList
java.util.HashSet
CacheableHashSet
java.util.LinkedHashSet
CacheableLinkedHashSet
119
Object Lifetimes
Pivotal GemFire Native Client
Object Lifetimes
The .NET API provides a managed set of assemblies for the C++ API. The underlying C++ object will stay
in memory until the .NET object is garbage-collected.
The underlying C++ API employs reference counting using smart pointers for most classes. This means
that all API operations with those objects return a reference to the underlying object and not a copy.
Consequently, the underlying object will not be freed as long as the .NET application holds a reference
to an object. In other words, the underlying object will stay in memory until the .NET object is garbagecollected. As long as a reference to an object is alive, the artifacts it maintains will also be alive.
For example, as long as a Region object is not garbage-collected, then the destructor of the C++ native
persistence manager (if any) for the region is not invoked.
In the C++ API, the references to an object are reduced when the object goes out of scope for stack
allocation, or is deleted explicitly for heap allocation. The object is destroyed when its reference count
reaches zero. In the .NET API, the references are reduced when the object is garbage-collected or is
explicitly disposed with the .NET using statement.
Because a reference to the objects is returned, any change to the object also immediately changes the
object as stored internally. For example, if an object is put into the cache using Region.Put , a reference
of the object is stored in the internal structures. If you modify the object, the internal object also changes.
However, it is not distributed to other members of the distributed system until another Region.Put is
done.
To find out if a class is reference counted, look at the online API documentation for the class. If the class is
wrapped by UMWrap or SBWrap, the class is reference counted.
These are examples of classes that are reference counted:
•
•
•
•
•
•
•
•
•
•
Cache
CacheStatistics
DistributedSystem
Properties
RegionAttributes
AttributesMutator
RegionEntry
Region
EntryEvent
RegionEvent
These are examples of classes that are not reference counted:
•
•
•
•
AttributesFactory
DataInput
DataOutput
SystemProperties
120
.NET Application Domains
Pivotal GemFire Native Client
.NET Application Domains
Application domains, or AppDomains, are units of isolation, security boundaries, and loading and
unloading for applications in the .NET runtime. Multiple application domains can run in a single process.
Each can have one or many threads, and a thread can switch application domains at runtime.
Note: Before you use application domains with GemFire, make sure you understand how to use
them generally in your .NET programming.
The .NET managed assemblies require interface methods invoked by the native C++ layer to be in the
same AppDomain as that of the .NET DLL. If not, an exception is thrown because the thread is unable to
cross AppDomain boundaries.
Problem Scenarios
These scenarios describe processes and implementations that should be avoided when using
AppDomains.
•
•
•
Using Application Callbacks on page 121
Loading an Application DLL in Multiple AppDomains on page 121
Native Client inside IIS on page 121
Using Application Callbacks
Scenario: A .NET thread loads the GemFire DLL in application domain AD1. This thread may have
access to the other domains in the application if code access security allows it. This thread can then
call AppDomain.CreateInstance to create a callback object ( ICacheListener, ICacheLoader,
or ICacheWriter) in another domain called AD2. If the callback object is marshaled by reference, the
callback is executed in the domain where it is created (AD2). The thread that loads the GemFire DLL in
domain AD1 runs the callback methods in the second domain, AD2. An exception is thrown when the
callback method is invoked because the native code that invokes the callback is not allowed to cross the
AppDomain boundary.
Resolution: When an application creates and unloads application domains it should ensure that the
application domain where the GemFire .NET DLL is loaded is the same domain where the application
callback and IGFSerializable objects are created.
Loading an Application DLL in Multiple AppDomains
Scenario: the application loads the GemFire DLL in one application domain, then reloads the GemFire
DLL in another application domain (with or without unloading the previous AppDomain ). The callbacks,
as well as other interface implementations, like IPdxSerializable and IGFSerializable, throw
exceptions because the native C++ code does not know about AppDomains and is loaded only once in the
initial AppDomain.
Resolution: The application should always use the first AppDomain to load the GemFire DLL, or it should
not load the GemFire DLL multiple times.
Native Client inside IIS
Scenario: When you deploy more than one web application inside an Internet Information Service (IIS),
the IIS creates an appdomain subprocess for each web application in the single process, but the native
client C++ cache instance remains a singleton in the process. Because of this, you can run into conflicts
between cache creation and closure by the different appdomains. For example, if one appdomain calls
121
Creating a Cache
Pivotal GemFire Native Client
cache.close, it closes the cache for the entire process. Any further cache access operations by the other
appdomains return cache closed exceptions.
Resolution: Cache create/close provides reference counting of Cache create and close. Each
process can use the counter to make sure it creates the Cache once and closes it once. To enable this, set
the GemFire system property, appdomain-enabled to true .
Creating a Cache
You create a cache using the GemFire CacheFactory.Create call. Cache creation initializes the
distributed system and creates the cache using your gfcpp.properties and cache.xml file settings
and any additional properties you provide to the call.
See Setting System and Cache Properties and see Cache Initialization File.
Connecting and Creating the Cache
In this example, the application connects to the distributed system and creates the cache
using the available configuration files.
The application becomes a distributed system member in the cache Create call.
CacheFactory cacheFactory = CacheFactory.CreateCacheFactory();
Cache cache = cacheFactory.Create();
Providing Properties to the Cache Creation
You can also create a cache by referencing a cache.xml file, as shown in the following
example. You can use the Properties object to change any of the gfcpp.properties
settings.
Properties prop = Properties.Create();
prop.Insert("cache-xml-file", "cache.xml");
CacheFactory cacheFactory =
CacheFactory.CreateCacheFactory(prop);
Cache cache = cacheFactory.Create();
For systems with security enabled, the credentials for a joining member are authenticated
when the cache is created and the system connection is made. For more information
about secure connections to a distributed system, see Security.
Creating a Region
To create a region, you create a RegionFactory using the RegionShortcut that most closely fits your
region configuration.
From that point, you customize the settings for region attributes as needed.
Creating a region using the native client API only creates a proxy client side region. A corresponding
region with the same name and path must also exist on the servers that have been configured for client
connections and upon which the client will perform its operations.
122
Adding an Entry to the Cache
Pivotal GemFire Native Client
Creating a CACHING_PROXY Region
This example creates a region using a CACHING_PROXY RegionShortcut with no further
modifications:
RegionFactory regionFactory =
cache.CreateRegionFactory(RegionShortcut.CACHING_PROXY);
IRegion<string, string> region = regionFactory
.Create<string, string>("exampleRegion");
Creating a CACHING_PROXY Region with LRU
This example creates a region based on the CACHING_PROXY RegionShortcut with two
additional region attributes settings. For information on the settings, see Region Attributes
Descriptions.
RegionFactory regionFactory =
cache.CreateRegionFactory(RegionShortcut.CACHING_PROXY);
// Create the example Region programmatically.
IRegion<string, string> region = regionFactory
.SetLruEntriesLimit(20000)
.SetInitialCapacity(20000)
.Create<string, string>("exampleRegion");
Adding an Entry to the Cache
You populate a native client region with cache entries by using the generic IDictionary API or by using
the .NET Region.Put or the Region.Create API functions.
The Put function places a new value into a region entry with the specified key, while the Create function
creates a new entry in the region. The Put and Create functions provide a user-defined parameter object
to any cache writer invoked in the process.
If a value for the entry key already exists in the cache when you add an entry, GemFire overwrites the
previously cached value. New values in the cache are propagated to the connected cache server.
The .NET Generics provide type safety, so you cannot change your entry key and value types once you
have begun to populate the region. If you need to use different types for the same region, store them all
inside objects in the region.
Using the API to Put Values Into the Cache
In this example, the program puts entries into the cache with string values.
region1["Key1"] = "Value1";
region1["Key2"] = "Value2";
region2["KeyA"] = 123;
region2["KeyB"] = 100;
region3.Put(111, "Value1", null);
region3.Put(222, "Value2", null);
123
Accessing an Entry
Pivotal GemFire Native Client
Using PutAll to Add Multiple Entries
You can batch up multiple key/value pairs into a hashmap and put them into the cache with a single
operation using the .NET Region.PutAll API function. Each entry is processed for interest registration
on the server, so each entry requires its own unique event ID. Updates and creates can be mixed in a
PutAll operation, so those events need to be addressed on the cache server for appropriate cache
listener invocation on distributed system members. Map entries retain their original order when they are
processed at the server.
Accessing an Entry
The region entry retrieval methods return the value associated with the specified key, and pass the
callback argument to any cache loaders or cache writers that are invoked during the operation.
If the value is not available locally, it is requested from the server. If the server request is unsuccessful, a
local cache loader is invoked, if one is available. The operation throws keyNotFoundException if the
Region is unable to retrieve a value through any of these means.
Using the Region API to Retrieve Values From the Cache
Here, the program retrieves two entries from the region.
string value1 = region1["Key1"];
string value2 = region1["Key2"];
int valueA = region2["KeyA"];
int valueB = region2["KeyB"];
string valueQ = region.Get(111, null);
string valueR = region.Get(222, null);
Data Serialization
All data that GemFire moves out of the local cache must be serializable.
Region data that must be serializable falls under the following categories:
•
•
•
•
•
•
•
Partitioned regions (except functions that add data locally to a partitioned region use the deserialized
form).
Distributed regions.
Regions that are persisted or overflowed to disk.
Server or client regions in a client/server installation.
Regions distributed between gateways in a multi-site installation.
Regions that receive events from remote caches.
Regions that provide function arguments and results.
To minimize the cost of serialization and deserialization, GemFire avoids changing the data format
whenever possible. This means your data may be stored in the cache in serialized or deserialized form,
depending on how you use it. For example, if a server acts only as a storage location for data distribution
between clients, it makes sense to leave the data in serialized form, ready to be transmitted to clients that
request it. Partitioned region data is always stored in serialized form with one exception—functions that
add data to a partitioned region locally use the deserialized form.
124
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
Data Serialization Options
Built-in .NET types are serialized automatically into the cache and can be retrieved by Java servers and
other GemFire clients. For domain objects that are not simple types, you have three GemFire serialization
options.
The options give good performance and flexibility for data storage, transfers, and language types. The
GemFire options can also improve performance in serializing and deserializing built-in types.
The simplest option is to use perform automatic serialization by registering the GemFire .NET PDX
reflection-based autoserializer in your application. When you have this registered, GemFire uses it for all
domain objects that are not custom serialized.
You can also custom serialize your objects by implementing one of the GemFire .NET
interfaces, GemStone::GemFire::Cache::Generic::IPdxSerializable or
GemStone::GemFire::Cache::IGFSerializable.
You also have the option of using default .NET serialization, but you cannot use it unless you also use
helper classes. The helper classes you must use are CacheableObject and CacheableObjectXml.
GemFire .NET PDX serialization has more bytes in overhead than GemFire .NET Data serialization, but
using PDX serialization helps you avoid the performance costs of deserialization when performing queries.
Applications can use PdxInstances in functions to avoid the deserialization of entire objects.
Table 18: Serialization Options: Comparison of Features
Capability
IGFSerializable
IPdxSerializable and PDX reflectionbased autoserializer
Handles multiple versions of
domain objects*
X
Provides single field access
on servers of serialized data,
without full deserialization.
Supported also for OQL
queries.
X
Automatically ported to other
languages by GemFire - no
need to program Java-side
implementation
X
Works with GemFire delta
propagation
X
X (See explanation below.)
*You can mix domain object versions where the differences between versions are the addition and removal
of object fields.
By default, you can use GemFire delta propagation with PDX serialization. However, delta propagation
will not work if you have set the GemFire property read-serialized to "true". In terms of deserialization, to
apply a change delta propagation requires a domain class instance and the fromDelta method. If you
have set read-serialized to true, you will receive an IPdxInstance instead of a domain class instance
and IPdxInstance does not have the fromDelta method required for delta propagation. You will also
require the Java domain class on the server similar to the you would need the .NET PDX Delta domain
class.
125
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
For detailed information on the interfaces, see the online API documentation.
Serialize with GemFire PDX Serialization
Pivotal GemFire's Portable Data eXchange (PDX) is a cross-language data format that can reduce the
cost of distributing and serializing your objects. PDX stores data in named fields that you can access
individually, to avoid the cost of deserializing the entire data object. PDX also allows you to mix versions of
objects where you have added or removed fields.
With the release of GemFire Native Client 7.0, PDX now works with C++ clients. If you have C++ clients
that may receive events for a region, you can now use PDX serialization on the region's entries.
You have two options for GemFire PDX serialization when using the C# .NET caching API. You can
program your domain objects using the IPdxSerializable interface or you can use GemFire’s
reflection-based autoserializer.
GemFire PDX Serialization Features
GemFire PDX serialization offers several advantages.
Application Versioning of PDX Domain Objects
Domain objects evolve along with your application code. You may create an address object with two
address lines, then realize later that a third line is required for some situations. Or you may realize that a
particular field is not used and want to get rid of it.
With PDX, you can use old and new versions of domain objects together in a distributed system if the
versions differ by the addition or removal of fields. This compatibility lets you gradually introduce modified
code and data into the system, without bringing the system down.
Pivotal GemFire maintains a central registry of the PDX domain object metadata. Using the registry,
GemFire preserves fields in each member's cache regardless of whether the member has the field
defined. When a member receives an object that has a field registered that the member is not aware of,
the member does not access the field, but preserves it and passes it along with the rest of the object to
other members. When a member receives an object that is missing one or more fields according to the
member's version, GemFire assigns the .NET default values for the field types to the missing fields.
Portability of PDX Serializable Objects
When you create an IPdxSerializable object, GemFire stores the object's type information in a central
registry. The information is passed between peers, between clients and servers, and between distributed
systems.
This offers a notable advantage to the .NET client, which shares data with Java cache servers. Clients
automatically pass registry information to servers when they store an IPdxSerializable object. Clients
can run queries and functions against the data in the servers without the servers needing to know anything
about the stored objects. One client can store data on the server to be retrieved by another client, with the
server never needing to know the object type. This means you can code your .NET clients to manage data
using Java servers without having to create Java implementations of your .NET domain objects.
Reduced Deserialization of Serialized Objects
The access methods for IPdxSerializable objects allow you to examine specific fields of your domain
object without deserializing the entire object. This can reduce serialization and deserialization costs
significantly. .NET clients can run queries and execute functions against the objects in the server caches
without deserializing the entire object on the server side. The query engine automatically recognizes PDX
objects and uses only the fields it needs.
126
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
Clients can execute Java functions on server data that only access parts of the domain objects by using
PdxInstance.
Likewise, peers can access just the fields needed from the serialized object, keeping the object stored in
the cache in serialized form.
Serialize Using the GemFire Pdx Autoserializer
When you register the reflection-based serializer, GemFire uses it to serialize all objects that do not
implement IPdxSerializable or IGFSerializable. You can customize the auto-serialization
behavior for your domain objects by adding serialization attributes to your object’s fields.
Procedure
1. If you have not already registered the PDX reflection-based autoserializer, add the registration code to
your application.
For example:
using GemStone.GemFire.Cache.Generic;
...
// Register reflection-based autoserializer to serialize
// domain objects using PDX serialization
Serializable.RegisterPdxSerializer(new ReflectionBasedAutoSerializer());
This can only be configured in the application code. It cannot be configured declaratively in
cache.xml.
2. (Optional) For each object you intend to have autoserialized, customize the serialization as needed.
Note: If you also use PDX serialization in Java for the object, customize your serialization the
same for both languages.
a. The following extension methods apply to autoserialization:
•
•
•
•
•
•
WriteTransform. Controls what field value is written during auto serialization.
ReadTransform. Controls what field value is read during auto deserialization.
GetFieldType. Defines the specific field names that will be generated during autoserialization.
IsIdentityField. Controls which field is marked as the identity field. Identity fields are used when
a PdxInstance computes its hash code to determine whether it is equal to another object.
GetFieldType. Determines the field type that will be used when autoserializing the given field.
IsFieldIncluded. Specifies which fields of a class to autoserialize.
See Extending the Autoserializer for sample usage.
b. If you are writing a Java application, you can use the IPdxType Mapper to map Java types to .NET
types. Note that you only need to use the IPdxTypeMapper if you are writing Java applications.
See Map .NET Domain Type Names to PDX Type Names with IPdxTypeMapper on page 134 for
sample usage.
c. To specify an identifier field in your domain object, add the attribute PdxIdentityField to the
field.
For example:
[PdxIdentityField] private int id;
d. To exclude a field from serialization, add the .NET attribute NonSerialized to the field.
For example:
[NonSerialized] private int myLocalData;
127
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
For each domain class GemFire serializes using the autoserializer, all fields are considered for serialization
except those defined as static, literal or readonly and those you explicitly exclude using the .NET
NonSerialized attribute.
Extend the PDX Autoserializer
This example code demonstrates how to extend the autoserializer to customize serialization.
Extending the Autoserializer
public class AutoSerializerEx : ReflectionBasedAutoSerializer
{
public override object WriteTransform(FieldInfo fi, Type
type, object originalValue)
{
if (fi.FieldType.Equals(Type.GetType("System.Guid")))
{
return originalValue.ToString();
}
else if
(fi.FieldType.Equals(Type.GetType("System.Decimal")))
{
return originalValue.ToString();
}
else
return base.WriteTransform(fi, type, originalValue);
}
public override object ReadTransform(FieldInfo fi, Type
type, object serializeValue)
{
if (fi.FieldType.Equals(Type.GetType("System.Guid")))
{
Guid g = new Guid((string)serializeValue);
return g;
}
else if
(fi.FieldType.Equals(Type.GetType("System.Decimal")))
{
return Convert.ToDecimal((string)serializeValue);
}
else
return base.ReadTransform(fi, type, serializeValue);
}
public override FieldType GetFieldType(FieldInfo fi, Type
type)
{
if (fi.FieldType.Equals(Type.GetType("System.Guid")) ||
fi.FieldType.Equals(Type.GetType("System.Decimal")))
return FieldType.STRING;
return base.GetFieldType(fi, type);
}
public override bool IsIdentityField(FieldInfo fi, Type type)
{
if (fi.Name == "_identityField")
return true;
return base.IsIdentityField(fi, type);
}
public override string GetFieldName(FieldInfo fi, Type type)
{
128
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
if (fi.Name == "_nameChange")
return fi.Name + "NewName";
}
}
return fi.Name ;
public override bool IsFieldIncluded(FieldInfo fi, Type type)
{
if (fi.Name == "_notInclude")
return false;
return base.IsFieldIncluded(fi, type);
}
Serialize Your Domain Objects with IPdxSerializer
For domain objects that you cannot or do not want to modify, use the IPdxSerializer class to serialize
and deserialize the object's fields.
You use one IPdxSerializer implementation for the entire cache, programming it for all of the domain
objects that you handle in this way. This way you do not have to implement the IPdxSerializable
interface for each domain class.
With IPdxSerializer, you leave your domain object as-is and handle the serialization and
deserialization in the separate serializer. You register the serializer in your cache PDX configuration. Then
program the serializer to handle all of the domain objects you need.
If you write your own IPdxSerializer and you also use the ReflectionBasedAutoSerializer,
then the IPdxSerializer needs to own the ReflectionBasedAutoSerializer and delegate to it. A
cache can only have a single IPdxSerializer instance.
Note: The IPdxSerializer toData and fromData methods differ from those for
IPdxSerializable. They have different parameters and results.
To register an IPdxSerializer, you can use the following code. Note that you can only register the
IPdxSerializer in the application code. It cannot be configured declaratively in cache.xml.
Example:
using GemStone.GemFire.Cache.Generic;
...
// Register a PdxSerializer to serialize
// domain objects using PDX serialization
Serializable.RegisterPdxSerializer(new MyPdxSerializer());
Serialize Using the GemFire IPdxSerializable Interface
Use this procedure to program your domain object for PDX serialization using the IPdxSerializable
Interface. When you write objects using PDX serialization, they are distributed to the server tier in PDX
serialized form. When you run queries against the objects on the servers, only the fields you specify are
deserialized.
Procedure
1. In your domain class, implement GemStone::GemFire::Cache::Generic::IPdxSerializable.
Example:
using GemStone.GemFire.Cache.Generic;
...
129
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
public class PortfolioPdx : IPdxSerializable
2. If your domain class does not have a zero-arg constructor, create one for it.
If you also use PDX serialization in Java for the object, serialize the object in the same way for each
language. Serialize the same fields in the same order and mark the same identify fields.
3. Program the IPdxSerializable ToData function to serialize your object as required by your
application.
a. Write your domain class's standard. NET data fields using the IPdxWriter write methods. GemFire
automatically provides IPdxWriter to the ToData function for IPdxSerializable objects.
b. Call the ToData markIdentifyField function for each field GemFire should use to identify
your object. This is used to compare objects for operations like DISTINCT queries. The
markIdentifyField call must come after the associated field write methods.
Example:
// object fields
private int m_id;
private string m_pkid;
private PositionPdx m_position1;
private PositionPdx m_position2;
private Hashtable m_positions;
private string m_type;
private string m_status;
private string[] m_names;
private byte[] m_newVal;
private DateTime m_creationDate;
private byte[] m_arrayZeroSize;
private byte[] m_arrayNull;
// ToData
public void ToData(IPdxWriter writer)
{
writer.WriteInt("id", m_id)
//identity field
.MarkIdentityField("id")
.WriteString("pkid", m_pkid)
.WriteObject("position1", m_position1)
.WriteObject("position2", m_position2)
.WriteObject("positions", m_positions)
.WriteString("type", m_type)
.WriteString("status", m_status)
.WriteStringArray("names", m_names)
.WriteByteArray("newVal", m_newVal)
.WriteDate("creationDate", m_creationDate)
.WriteByteArray("arrayNull", m_arrayNull)
.WriteByteArray("arrayZeroSize", m_arrayZeroSize);
}
4. Program IPdxSerializable FromData to read your data fields from the serialized form into the
object's fields using the IPdxReader read methods. GemFire automatically provides IPdxReader to
the FromData function for IPdxSerializable objects.
Use the same names as you did in ToData and call the read operations in the same order as you
called the write operations in your ToData implementation.
Example:
public void FromData(IPdxReader reader)
{
m_id = reader.ReadInt("id");
bool isIdentity = reader.IsIdentityField("id");
130
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
if (isIdentity == false)
throw new IllegalStateException("Portfolio id is identity
field");
bool isId = reader.HasField("id");
if (isId == false)
throw new IllegalStateException("Portfolio id field not found");
bool isNotId = reader.HasField("ID");
if (isNotId == true)
throw new IllegalStateException("Portfolio isNotId field found");
m_pkid = reader.ReadString("pkid");
m_position1 = (PositionPdx)reader.ReadObject("position1");
m_position2 = (PositionPdx)reader.ReadObject("position2");
m_positions = (Hashtable)reader.ReadObject("positions");
m_type = reader.ReadString("type");
m_status = reader.ReadString("status");
m_names = reader.ReadStringArray("names");
m_newVal = reader.ReadByteArray("newVal");
m_creationDate = reader.ReadDate("creationDate");
m_arrayNull = reader.ReadByteArray("arrayNull");
m_arrayZeroSize = reader.ReadByteArray("arrayZeroSize");
}
5. Optionally, program your domain object's equals and hashcode methods.
Program Your Application to Use IPdxInstance
An IPdxInstance is a lightweight wrapper around PDX serialized bytes. It provides applications with runtime access to fields of a PDX serialized object.
You can configure your cache to return an IPdxInstance when a PDX serialized object is deserialized
instead of deserializing the object to a domain class. You can then program your application code that
reads your entries to handle IPdxInstances fetched from the cache.
Note: This option applies only to entry retrieval that you explicitly code using methods like
EntryEvent.getNewValue and Region.get, as you do inside functions or in cache listener
code. This does not apply to querying because the query engine retrieves the entries and handles
object access for you.
Note: IPdxInstance overrides any custom implementation you might have coded for your
object's equals and hashcode methods.
Procedure
1. In the cache.xml file of the server member where entry fetches are run, set the <pdx> read-serialized
attribute to true.
Data is not necessarily accessed on the member that you have coded for it. For example, if a client
application runs a function on a server, the actual data access is done on the server, so you set readserialized to true on the server.
For example:
// Cache configuration setting PDX read behavior
<cache>
<pdx read-serialized="true" />
... </cache>
131
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
2. Write the application code that fetches data from the cache to handle a IPdxInstance. If you are sure
you will only retrieve IPdxInstances from the cache, you can code only for that. In many cases, a
IPdxInstance or a domain object may be returned from your cache entry retrieval operation, so you
should check the object type and handle each possible type.
See Creating an IPdxInstance with IPdxInstanceFactory on page 132 for an example of this.
If you configure your cache to allow PDX serialized reads, cache fetches return the data in the form it is
found. If the object is not serialized, the fetch returns the domain object. If the object is serialized, the fetch
returns the PdxInstance for the object.
Note: If you are using IPdxInstances, you cannot use delta propagation to apply changes to
PDX serialized objects.
For example, in client/server applications that are programmed and configured to handle all data activity
from the client, PDX serialized reads done on the server side will always return the IPdxInstance. This
is because all of data is serialized for transfer from the client and you are not performing any server-side
activities that would deserialize the objects in the server cache.
In mixed situations, such as where a server cache is populated from client operations and also from data
loads done on the server side, fetches done on the server can return a mix of IPdxInstances and
domain objects.
When fetching data in a cache with PDX serialized reads enabled, the safest approach is to code to handle
both types, receiving an Object from the fetch operation, checking the type and casting as appropriate.
Use the IPdxInstanceFactory to Create IPdxInstances
You can use the IPdxInstanceFactory to create an IPdxInstance from raw data when the domain
class is not available on the server.
This option can be useful when you need an instance of a domain class for plug-in code such as a function
or a loader. If you have the raw data for the domain object (the class name and each field's type and
data), then you can explicitly create a IPdxInstance. The IPdxInstanceFactory is very similar to
the IPdxWriter except that after writing each field, you need to call the create method which returns the
created IPdxInstance.
Creating an IPdxInstance with IPdxInstanceFactory
/*
* The PdxInstance QuickStart Example.
* This example takes the following steps:
*
* This example shows IPdxInstanceFactory and
IPdxInstance usage.
*
* 1. Create a GemFire Cache.
* 2. Creates the PdxInstanceFactory for Person class.
* 3. Then creates instance of PdxInstance
* 4. It does put.
* 5. Then it does get and access it fields.
* 6. Close the Cache.
*
*/
// Use standard namespaces
using System;
using System.Reflection;
// Use the GemFire namespace
using GemStone.GemFire.Cache.Generic;
namespace GemStone.GemFire.Cache.Generic.QuickStart
{
public class Person
132
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
{
private string name;
//this is the only field used on server to create
hashcode and use in equals method
[PdxIdentityField]
private int id;
private int age;
public Person() { }
public Person(string name, int id, int age)
{
this.name = name;
this.id = id;
this.age = age;
}
#region Public Properties
public string Name
{
get { return name; }
}
public int ID
{
get { return id; }
}
public int Age
{
get { return age; }
}
#endregion
}
// The PdxRemoteQuery QuickStart example.
class PdxInstance
{
static void Main(string[] args)
{
try {
CacheFactory cacheFactory =
CacheFactory.CreateCacheFactory(null);
Console.WriteLine("Connected to the GemFire Distributed
System");
// Create a GemFire Cache with the
"clientPdxRemoteQuery.xml" Cache XML file.
// Set SetPdxReadSerialized to true to access
PdxInstance
Cache cache = cacheFactory.Set("cache-xml-file",
"XMLs/clientPdxInstance.xml").Create();
Console.WriteLine("Created the GemFire Cache");
// Get the example Region from the Cache which is
declared in the
Cache XML file.
IRegion<string, IPdxInstance> region =
cache.GetRegion<string,
IPdxInstance>("Person");
Console.WriteLine("Obtained the Region from the Cache");
Person p = new Person("Jack", 7, 21);
//PdxInstanceFactory for Person class
IPdxInstanceFactory pif =
cache.CreatePdxInstanceFactory("Person");
pif.WriteString("name", p.Name);
pif.WriteInt("id", p.ID);
pif.MarkIdentityField("id");
pif.WriteInt("age", p.Age);
IPdxInstance pdxInstance = pif.Create();
Console.WriteLine("Created PdxInstance for Person
class");
133
Pivotal GemFire Native Client C# .NET API
p.Name
true)
Pivotal GemFire Native Client
region["Key1"] = pdxInstance;
Console.WriteLine("Populated PdxInstance Object");
IPdxInstance retPdxInstance = region["Key1"];
if((int)retPdxInstance.GetField("id") == p.ID
&& (int)retPdxInstance.GetField("age") == p.Age
&& (string)retPdxInstance.GetField("name") ==
&& retPdxInstance.IsIdentityField("id") ==
Console.WriteLine("PdxInstance returns all fields
value
expected");
else
Console.WriteLine("PdxInstance doesn't returns all
fields value
expected");
// Close the GemFire Cache.
cache.Close();
Console.WriteLine("Closed the GemFire Cache");
}
// An exception should not occur
catch (GemFireException gfex)
{
Console.WriteLine("PdxSerializer GemFire Exception:
{0}",
gfex.Message);
}
}
}
}
Map .NET Domain Type Names to PDX Type Names with
IPdxTypeMapper
PDX serialized instances in Java map to .NET types with the same name. If you need to adjust the .NET
name, then you need to use the IPdxTypeMapper.
See the Java to .NET Type Mapping Table for current mappings.
Using IPdxTypeMapper
//This demonstrates, how to map .NET type to pdx type or java
type
public class PdxTypeMapper : IPdxTypeMapper
{
public string ToPdxTypeName(string localTypeName)
{
return "pdx_" + localTypeName;
}
public string FromPdxTypeName(string pdxTypeName)
{
return pdxTypeName.Substring(4);//need to
extract "pdx_"
}
}
134
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
Serialize with the GemFire IGFSerializable Interface
The .NET IGFSerializable interface provides fast and compact data serialization.
Generic and Custom Serializable Types
All built-in generics are automatically registered at initialization. You have a couple of options for complex
key types.
If your application uses more complex key types that you want to make more accessible or easier to
handle, you can derive a new class from IGFSerializable. Another option is for the application to do its
own object serialization using Byte[] or a custom type.
Blobs
If you have data that is best handled as a blob, such as structs that do not contain pointers, use a
Byte[] or, if you need something more complex than Byte[], implement a custom type using either
IPdxSerializable or IGFSerializable.
Object Graphs
If you have a graph of objects in which each node can be serializable, the parent node calls
DataOutput.WriteObject to delegate the serialization responsibility to its child nodes. Similarly, your
application calls DataInput.ReadObject to deserialize the object graph.
Note: The GemFire IGFSerializable interface does not support object graphs with multiple
references to the same object. If your application uses these types of circular graphs, you must
address this design concern explicitly.
For more information, see the online API documentation for DataOutput and DataInput.
How Serialization Works with IGFSerializable
When your application puts an object into the cache for distribution, Pivotal GemFire serializes the data by
taking these steps.
1. Calls the appropriate ClassId function and creates the TypeId from it.
2. Writes the TypeId for the instance.
3. Invokes the ToData function for the instance.
When your application subsequently receives a byte array, GemFire take the following steps:
1. Decodes the TypeId and creates an object of the designated type, using the registered factory
functions.
2. Invokes the FromData function with input from the data stream.
3. Decodes the data and then populates the data fields.
The TypeId is an integer of four bytes, which is a combination of ClassId integer and 0x27, which is an
indicator of user-defined type.
Implement the IGFSerializable Interface
To store your own data types in the cache, you implement the GemFire IGFSerializable interface.
Examples follow the procedure.
Procedure
135
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
1. Implement the ToData function that serializes your data:
void ToData(DataOutput output)
The ToData function is responsible for copying all of the data fields for the object to the object stream.
The DataOutput class represents the output stream and provides methods for writing the primitives in
a network byte order. For more about this, see the online API documentation for DataOutput.
2. Implement the FromData function that consumes a data input stream and repopulates the data fields
for the object:
void fromData (DataInput& input)
The DataInput class represents the input stream and provides methods for reading input elements.
The FromData function must read the elements of the input stream in the same order that they were
written by ToData. For more about this, see the online API documentation for DataInput.
3. Implement the ClassId function to return an integer which is unique for your class (in the set of all of
your user-defined classes).
Simple BankAccount Class
This example shows a simple class, BankAccount, that encapsulates two ints:
customerId and accountId:
public class BankAccount
{
private int m_customerId;
private int m_accountId;
public int Customer
{
get
{
return m_customerId;
}
}
public int Account
{
get
{
return m_accountId;
}
}
public BankAccount(int customer, int account)
{
m_customerId = customer;
m_accountId = account;
}
}
Implementing a Serializable Class
To make BankAccount serializable, you implement the IGFSerializable interface as
shown in this example:
public class BankAccount : IGFSerializable
{
private int m_customerId;
private int m_accountId;
public int Customer
{
136
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
get
{
}
}
return m_customerId;
}
public int Account
{
get
{
return m_accountId;
}
}
public BankAccount(int customer, int account)
{
m_customerId = customer;
m_accountId = account;
}
// Our TypeFactoryMethod
public static IGFSerializable CreateInstance()
{
return new BankAccount(0, 0);
}
#region IGFSerializable Members
public void ToData(DataOutput output)
{
output.WriteInt32(m_customerId);
output.WriteInt32(m_accountId);
}
public IGFSerializable FromData(DataInput input)
{
m_customerId = input.ReadInt32();
m_accountId = input.ReadInt32();
return this;
}
public UInt32 ClassId
{
get
{
return 11;
}
}
public UInt32 ObjectSize
{
get
{
return (UInt32)(sizeof(Int32) + sizeof(Int32));
}
}
Register the Type
To use the BankAccount type, you must register it with the type system. Then, when an incoming stream
contains a BankAccount, it can be manufactured from the associated TypeFactoryMethod.
Serializable.RegisterType(BankAccount.CreateInstance);
Typically, you would register the type before calling the function DistributedSystem.Connect.
137
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
Using ClassId
A ClassId is an integer that returns the ClassId of the instance being serialized. The ClassId is used
by deserialization to determine what instance type to create and deserialize into.
Using DSFID
A DSFID is an integer that returns the data serialization fixed ID type. DSFID is used to determine what
instance type to create and deserialize into. DSFID should not be overridden by custom implementations,
and it is reserved only for built-in serializable types.
Using Custom Key Types
If your application uses its own key types that are too complex to easily force into string, you can probably
improve performance by using a custom type and implementing HashCode and Equals functions. For
example, if you have hybrid data types such as floating point numbers, you can implement your own type
that encapsulates the floating point number. Comparing floating point numbers in this way provides greater
performance than comparing a string representation of the floating point numbers, with such noticeable
improvements as faster cache access and smaller payloads.
See Serialization in Native Client Mode with a Java Server on page 106 for information about implementing
key types for a native client that is used with a Java cache server.
To extend a type that implements IPdxSerializable or IGFSerializable for your key, override and
implement the HashCode and Equals methods in the key as needed.
Using a Custom Class With IGFSerializable
An example shows how to use the BankAccount custom key type and the AccountHistory value type
that were previously defined.
Using a BankAccount Object
class AccountHistory : IGFSerializable
{
#region Private members
private List<string> m_history;
#endregion
public AccountHistory()
{
m_history = new List<string>();
}
public void ShowAccountHistory()
{
Console.WriteLine("AccountHistory:");
foreach (string hist in m_history) {
Console.WriteLine("\t{0}", hist);
}
}
public void AddLog(string entry)
{
m_history.Add(entry);
}
public static IGFSerializable CreateInstance()
{
return new AccountHistory();
}
#region IGFSerializable Members
public IGFSerializable FromData(DataInput input)
{
138
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
int len = input.ReadInt32();
m_history.Clear();
for (int i = 0; i < len; i++) {
m_history.Add(input.ReadUTF());
}
return this;
}
public void ToData(DataOutput output)
{
output.WriteInt32(m_history.Count);
foreach (string hist in m_history) {
output.WriteUTF(hist);
}
}
public UInt32 ClassId
{
get
{
return 0x05;
}
}
public UInt32 ObjectSize
{
get
{
UInt32 objectSize = 0;
foreach (string hist in m_history) {
objectSize += (UInt32)(hist == null ? 0 :
sizeof(char) * hist.Length);
}
return objectSize;
}
}
#endregion
}
public class TestBankAccount
{
public static void Main()
{
// Register the user-defined serializable type.
Serializable.RegisterType(AccountHistory.CreateInstance);
Serializable.RegisterType(BankAccountKey.CreateInstance);
// Create a cache.
CacheFactory cacheFactory =
CacheFactory.CreateCacheFactory(null);
Cache cache = cacheFactory.Create();
// Create a region.
RegionFactory regionFactory =
cache.CreateRegionFactory(RegionShortcut.CACHING_PROXY);
Region region =
regionFactory.Create("BankAccounts");
// Place some instances of BankAccount cache region.
BankAccountKey baKey = new BankAccountKey(2309,
123091);
AccountHistory ahVal = new AccountHistory();
ahVal.AddLog("Created account");
region.Put(baKey, ahVal);
Console.WriteLine("Put an AccountHistory in cache
keyed with
BankAccount.");
// Display the BankAccount information.
139
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
Console.WriteLine(baKey.ToString());
// Call custom behavior on instance of
AccountHistory.
ahVal.ShowAccountHistory();
// Get a value out of the region.
AccountHistory history = region.Get(baKey) as
AccountHistory;
if (history != null)
{
Console.WriteLine("Found AccountHistory in the
cache.");
history.ShowAccountHistory();
history.AddLog("debit $1,000,000.");
region.Put(baKey, history);
Console.WriteLine("Updated AccountHistory in the
cache.");
}
// Look up the history again.
history = region.Get(baKey) as AccountHistory;
if (history != null)
{
Console.WriteLine("Found AccountHistory in the
cache.");
history.ShowAccountHistory();
}
// Close the cache.
cache.Close();
}
}
//Example 5.12 Using ICacheLoader to Load New Integers in
the Region
class ExampleLoaderCallback : ICacheLoader
{
#region Private members
private int m_loads = 0;
#endregion
#region Public accessors
public int Loads
{
get
{
return m_loads;
}
}
#endregion
}
Application Callbacks
For region-level events, an application can use AttributesFactory.SetCache* methods to implement
and register the ICacheLoader, ICacheWriter, and ICacheListener interfaces to perform custom
actions.
You can use Region.Put for simple caching situations. For more complex needs, you should implement
the ICacheLoader interface and allow the cache to manage the creation and loading of objects. When a
Region.Get is called for a region entry with a value of null, the ICacheLoader::Load method of the
cache loader (if any) for the region is invoked. A static CacheLoader::NetSearch method is provided
which can be used by ICacheLoader implementations to locate the requested key in the distributed
system. The ICacheListener interface can be used to listen to various region events after events such
140
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
as create, update, or invalidate of region entries have occurred. The ICacheWriter interface is invoked
before the events have occurred.
Using ICacheLoader to Load New Integers in the Region
This example demonstrates an ICacheLoader implementation for loading new integers
into a region.
class SimpleCacheLoader<TKey, TVal> : ICacheLoader<TKey, TVal>
{
#region ICacheLoader Members
public TVal Load(IRegion<TKey, TVal> region, TKey key,
object helper)
{
Console.WriteLine("SimpleCacheLoader: Received Load
event for region:
{0} and key: {1}", region.Name, key);
return default(TVal);
}
public void Close(IRegion<TKey, TVal> region)
{
Console.WriteLine("SimpleCacheLoader: Received Close
event of region:
{0}", region.Name);
}
#endregion
}
Using ICacheWriter to Track Creates and Updates for a Region
This example implements ICacheWriter to track region entry create and update
events. This example just reports the events to the screen, but you can do whatever you
need to do with the events.
class SimpleCacheWriter<TKey, TVal> : ICacheWriter<TKey, TVal>
{
#region ICacheWriter<TKey, TVal> Members
public bool BeforeUpdate(EntryEvent<TKey, TVal> ev)
{
Console.WriteLine("SimpleCacheWriter: Received
BeforeUpdate event for: {0}", ev.Key);
return true;
}
// ... handle other entry events as needed
public bool BeforeRegionClear(RegionEvent<TKey, TVal> ev)
{
Console.WriteLine("SimpleCacheWriter: Received
BeforeRegionClear event of region: {0}",
ev.Region.Name);
return true;
}
// ... handle other region events as needed
#endregion
}
141
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
A Sample ICacheListener Implementation
This example implements ICacheListener.
class SimpleCacheListener<TKey, TVal> : ICacheListener<TKey,
TVal>
{
#region ICacheListener<TKey, TVal> Members
public void AfterCreate(EntryEvent<TKey, TVal> ev)
{
Console.WriteLine("SimpleCacheListener: Received
AfterCreate event
for: {0}", ev.Key);
}
// ... handle other entry events as needed
public void AfterRegionDestroy(RegionEvent<TKey, TVal> ev)
{
Console.WriteLine("SimpleCacheListener: Received
AfterRegionDestroy
event of region: {0}", ev.Region.Name);
}
// ... handle other region events as needed
#endregion
}
A Simple C# Example
An example shows how to connect to GemFire, create a cache and region, put and get keys and values,
and disconnect.
Simple C# Code
class BasicOperations
{
public static void Main(string[] args)
{
try
{
// 1. Create a cache
CacheFactory cacheFactory =
CacheFactory.CreateCacheFactory();
Cache cache = cacheFactory.Create();
factory
// 2. Create default region attributes using region
RegionFactory regionFactory =
cache.CreateRegionFactory(RegionShortcut.CACHING_PROXY);
// 3. Create a region
IRegion<int, string> region =
regionFactory.Create<int, string>("exampleRegion");
// 4. Put some entries
region[111] = "Value1";
region[123] = "123";
// 5. Get the entries
string result1 = region[111];
string result2 = region[123];
142
Troubleshooting .NET Applications
}
}
}
Pivotal GemFire Native Client
// 6. Close cache
cache.Close();
Troubleshooting .NET Applications
The .NET Framework does not find managed DLLs using the conventional PATH environment variable. In
order for your assembly to find and load a managed DLL, it must either be loaded as a private assembly
using assemblyBinding, or it must be installed into the Global Assembly Cache (GAC).
The .NET Framework does not find managed DLLs using the conventional PATH environment variable. In
order for your assembly to find and load a managed DLL, it must either be loaded as a private assembly
using assemblyBinding, or it must be installed into the Global Assembly Cache (GAC).
The GAC utility must be run on every machine that runs the .NET code.
If an assembly attempts to load the GemStone.GemFire.Cache.dll without meeting this requirement,
you receive this System.IO.FileNotFoundException:
{{
Unhandled Exception: System.IO.FileNotFoundException: Could not load file
or assembly 'GemStone.GemFire.Cache, Version=7.0.0.0, Culture=neutral,
PublicKeyToken= 126e6338d9f55e0c' or one of its dependencies. The system
cannot find the file specified.
File name: 'GemStone.GemFire.Cache, Version=7.0.0.0, Culture=neutral,
PublicKeyT oken=126e6338d9f55e0c'
at HierarchicalClient.Main()
}}
Resolving the Error
Each computer where the common language runtime is installed has a machine-wide code cache called
the Global Assembly Cache (GAC). The global assembly cache stores assemblies specifically designated
to be shared by several applications on the computer.
As a general guideline, keep assembly dependencies private, and locate assemblies in the application
directory unless sharing an assembly is explicitly required. Share assemblies by installing them into the
global assembly cache only when necessary.
Using GemStone.GemFire.Cache.dll As a Private
Assembly
To access GemStone.GemFire.Cache.dll as a private assembly, you need to specify a .config file
for your application.
The file needs to be the same name as your application, with a .config suffix. For example, the
.config file for main.exe would be main.exe.config. The two files must reside in the same directory.
Follow these steps to create a .config file:
143
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
1. Copy %GFCPP%/docs/default.exe.config to the appropriate location.
2. Rename default.exe.config to the name of your application.
3. Change the href attribute of the CodeBase element to point to your
GemStone.GemFire.Cache.dll file. Either http, relative, or absolute paths will work.
A Sample .config File
The following example shows an excerpt of a .config file. The PublicKeyToken value
is only an example, and the codebase version value is not set correctly. See %GFCPP%/
docs/default.exe.config for an actual example for this release.
<configuration>
<runtime>
<assemblyBinding
xmlns="urn:schemas-microsoft-com:asm.v1">
<dependentAssembly>
<assemblyIdentity name="GemStone.GemFire.Cache"
publicKeyToken="126e6338d9f55e0c"
culture="neutral" />
<codeBase version="0.0.0.0"
href="../../bin/GemStone.GemFire.Cache.dll"/>
</dependentAssembly>
</assemblyBinding>
</runtime>
</configuration>
Note: If the .config file contain errors, no warning or error messages are
issued. The application runs as if no .config file is present.
Implementing the Shared Assembly
Follow these steps to install the shared assembly into the Global Assembly Cache (GAC).
1. Go to the NativeClient_xxxx directory.
cd %GFCPP%
2. Run the GAC utility to install GemStone.GemFire.Cache.dll into the GAC.
gacutil.exe /if GemStone.GemFire.Cache.dll
When you are ready to uninstall, use the /u switch. More information on the GAC utility can be found at
http://www.msdn.com, or by using gacutil.exe /?.
144
Pivotal GemFire Native Client C# .NET API
Pivotal GemFire Native Client
145
Preserving Data
Pivotal GemFire Native Client
Part
VIII
Preserving Data
Topics:
•
•
•
High Availability for ClientServer Communication
Enabling Queue Conflation to
Improve Update Performance
Durable Client Messaging
Preserving Data explains how to configure the Pivotal GemFire native
client to preserve data when the cache server loses its connection
to the client, or when the native client is temporarily disconnected or
abnormally terminated from the distributed system.
Preserving Dataexplains how to configure the Pivotal GemFire native
client to preserve data when the cache server loses its connection
to the client, or when the native client is temporarily disconnected or
abnormally terminated from the distributed system.
Data preservation is accomplished by ensuring that reliable event
messaging is maintained between the client and the cache server.
Any lost messages intended to be sent to a client as create , update or
invalidate events for cached entries can quickly cause the data in the
client cache to be unsynchronized and out of date with the rest of the
distributed system.
146
High Availability for Client-Server Communication
Pivotal GemFire Native Client
High Availability for Client-Server Communication
The GemFire native client provides reliable event messaging from cache server to client to prevent
data loss during server failover operations. High availability is implemented in the cache server and is
configured in the native client.
The GemFire native client provides reliable event messaging from cache server to client to prevent
data loss during server failover operations. High availability is implemented in the cache server and is
configured in the native client.
See Configuring Highly Available Servers in the GemFire User's Guide for details about configuring a Java
cache server for high availability.
Note: High availability functions the same whether the region is partitioned or not. See Partitioned
Regions in the Pivotal GemFire User's Guide for information about partitioned regions.
Configuring Native Clients for High Availability
Configure high availability by setting the pool attribute subscription-redundancy to the number of
copies you want maintained.
A client maintains its queue redundancy level at the time of a primary server failure by connecting to
additional secondary servers.
Native clients can specify the number of secondary servers where the client registers interest and
maintains subscription channels, in addition to the subscription channel with the primary server. The
secondary servers maintain redundant update queues for the client. If the primary server fails, a secondary
becomes a primary to provide uninterrupted messaging to the client. If possible, another secondary is then
initialized so the total number of secondaries is not reduced by the failover.
Setting the Server Redundancy Level in cache.xml
This example sets one redundant server as failover backup to the primary server:
<cache>
<pool name="examplePool"
subscription-enabled="true" subscription-redundancy="1">
<server host="java_servername1" port="java_port1" />
<server host="java_servername2" port="java_port2" />
</pool>
<region name = "ThinClientRegion1" >
<region-attributes refid="CACHING_PROXY" poolname="examplePool"/>
</region>
</cache>
Setting the Server Redundancy Level Programmatically
You can set the redundancy level programmatically. This example creates a client cache
with two redundant cache servers configured in addition to the primary server.
The server redundancy level can be configured using the pool API. For more information
about the pool API, see Using Connection Pools.
PropertiesPtr pp = Properties::create( );
systemPtr = CacheFactory::createCacheFactory(pp);
147
Enabling Queue Conflation to Improve Update Performance
Pivotal GemFire Native Client
// Create a cache.
cachePtr = systemPtr->setSubscriptionEnabled(true)
->addServer("localhost", 24680)
->addServer("localhost", 24681)
->addServer("localhost", 24682)
->setSubscriptionRedundancy(2)
->create();
When failover to a secondary server occurs, a new secondary is added to the redundancy
set. If no new secondary server is found, the redundancy level is not satisfied but the
failover procedure completes successfully. Any new live server is added as a secondary
and interest is registered on it.
Sending Periodic Acknowledgment
Servers use periodic acknowledgment to reduce the likelihood of duplicate notifications, and to reduce
resource usage.
When redundancy is enabled for high availability and redundancy-level is set to 1 or higher, clients
send (and servers expect) periodic acknowledgment messages at configurable intervals for notifications
they have received. A periodic ack is not sent by the client if there are no unacknowledged notifications at
the time.
Use the following system properties in the gfcpp.properties file to configure periodic ack.
notify-ack-interval
notify-dupcheck-life
Minimum period between two consecutive
acknowledgment messages sent from the client to
the server. The default setting (in seconds) is 10.
Minimum time a client continues to track a
notification source for duplicates when no new
notifications arrive before expiring it. The default
setting (in seconds) is 300.
The Pool API also provides attributes to configure periodic ack and duplicate message tracking timeout.
See subscription-message-tracking-timeout and subscription-ack-interval in the list of
pool attributes under Configuring Pools for Servers or Locators.
Enabling Queue Conflation to Improve Update
Performance
You can configure client queues on the cache server to conflate queued messages, so the native client
receives only the latest update for a particular entry key.
You can configure client queues on the cache server to conflate queued messages, so the native client
receives only the latest update for a particular entry key.
You enable conflation in the cache server region, so all clients interested in updates in a particular region
either get the updates conflated or not. To enable conflation, set the cache server’s enable-bridgeconflation region attribute to true. Conflation is set to false by default.
Queue conflation is performed when an entry update is added to the queue. If the last operation queued for
that key is also an update operation, the previously enqueued update is removed, leaving only the latest
148
Durable Client Messaging
Pivotal GemFire Native Client
update to be sent to the client when event distribution occurs. For high availability, the conflation process is
also performed on any secondary queues.
Only entry update messages in a cache server region with distributed-no-ack scope are conflated.
Region operations and entry operations other than updates are not conflated.
For more information, see Conflate the Server Subscription Queue in the GemFire User’s Guide.
Overriding Queue Conflation Per-Client
You can override conflation on a per-client basis by setting the conflate-events property in the native
client’s gfcpp.properties file.
Valid settings are:
•
•
•
server. Uses the server settings.
true. Conflates everything sent to the client.
false.Does not conflate anything sent to the client.
Durable Client Messaging
You can configure the redundancy level for client queues that are stored on cache servers. This ensures
that the client will not lose messages if it loses the connection to its primary server.
You can configure the redundancy level for client queues that are stored on cache servers. This ensures
that the client will not lose messages if it loses the connection to its primary server.
Durable messaging allows a disconnected client application to recover its subscribed data when it
reconnects to the cache server because the server continues to queue messages for which the client has
registered interest.
Durable Client Messaging Requirements
The messaging queues used for durable messaging are the same regular messaging queues used for
basic server-to-client messaging, with additional requirements.
Requirements, options, and functionality of messaging queues are described in the GemFire User’s Guide.
If you are using highly available servers, see High Availability for Client-Server Communication on page
147for additional requirements.
For durable client messaging, you also need the following:
•
•
•
•
Durable clients. If the client is durable, the server continues to maintain the client queues when the
client disconnects.
Note: Redundancy management is handled by the client, so when the client is disconnected
from the server the redundancy of client events is not maintained. Even if the servers fail one at
a time, so that running clients have time to fail over and pick new secondary servers, an offline
durable client cannot fail over. As a result, the client loses its queued messages.
Durable interest registration. A durable client’s interest registrations specify whether its interest
in a key is durable. If it is, the servers continue queuing messages for that key while the client is
disconnected.
Cache ready message. When it is ready to receive the stored messages, a durable client must call
Cache.readyForEvents to send a cache ready message to the server.
Disconnect keepalive specification. When a durable client disconnects normally it must tell the server
whether to maintain the message queue or delete it.
149
Preserving Data
•
Pivotal GemFire Native Client
Durable client callback method. If you use cache listeners on the durable clients, you have the option
to implement the afterRegionLive callback method. This callback is invoked after the durable client
connects to its servers, when it has received all of its stored messages and replayed the events.
Client-Side Configuration
All durable messaging configurations are performed on the client.
Configuring a Durable Native Client
The durable native client can be configured in the gfcpp.properties file, or in the
CacheFactory::set(name, value) call.
•
•
Durable client ID—You indicate that the client is durable by giving it a durable-client-ID. The
servers use this ID to identify the client. For a non-durable client, the durable-client-ID is an
empty string. The ID can be any number that is unique among the clients attached to servers in the
same distributed system.
Durable timeout—The durable-timeout setting specifies how long this client’s servers should
wait after the client disconnects before terminating its message queue. During that time, the servers
consider the client alive and continue to accumulate messages for it. The default is 300 seconds.
The durable-timeout setting is a tuning parameter. When setting the timeout, take into account the
normal activity of your application, the average size of your messages, and the level of risk you can handle.
Assuming that no messages are being removed from the queue, how long can the application run before
the queue reaches the maximum message count? In addition, how long can it run before the queued
messages consume all the memory on the client host? How serious is each of those failures to your
operation?
To assist with tuning, GemFire provides statistics that track message queues for durable clients through
the disconnect and reconnect cycles. For statistics documentation, see Statisticsin the GemFire User's
Guide.
When the queue is full, it blocks further operations that add messages until the queue size drops to an
acceptable level. The action to take is specified on the server. For details on configuring the queue, see
Implementing Durable Client/Server Messaging in the GemFire User's Guide.
Configuring a Durable Native Client Using gfcpp.properties
The following example shows gfcpp.properties settings to make the client durable
and set the durable timeout to 200seconds.
durable-client-id=31
durable-timeout=200
Configuring a Durable Client Through the API (C++)
This programmatic example creates a durable client using the
CacheFactory::set(name, value).
// Create durable client's properties using the C++ api
PropertiesPtr pp = Properties::create();
pp->insert("durable-client-id", "DurableClientId");
pp->insert("durable-timeout", 200);
cacheFactoryPtr = CacheFactory::createCacheFactory(pp);
150
Preserving Data
Pivotal GemFire Native Client
Configuring Durable Interest in Keys
When a durable client disconnects, its servers keep queuing messages for selected keys. The client
indicates which keys by registering durable interest for those keys.
This fine-grained control handles the constraints of queue size and memory by saving only the critical
messages.
You still register interest for other keys, but not durable interest. When the client is connected to its servers,
it receives messages for those non-durable keys. When the client is disconnected, its non-durable interest
registrations are deleted but messages that are already in the queue remain there.
For durable clients, all interest registration is done immediately after the regions are created. This is
required whether interest registration is durable or not durable. An extra registerInterest parameter
specified for durable clients indicates whether the registration is durable ( true ) or not ( false ).
API Client Durable Interest List Registration (C++)
The following programmatic example registers durable interest in Key-1 . The interest
registration happens immediately after region creation and before anything else.
// Durable client interest registration can be
// durable (true) or nondurable(default).
VectorOfCacheableKey keys;
keys.push_back( CacheableString::create("Key-1") );
regionPtr->registerKeys(keys,true);
You use the typical methods for interest registration and configure notification by subscription on the server
as usual. For details, see Registering Interest for Entries.
Note: Changing interest registration after the durable client connects the first time can cause data
inconsistency and is not recommended.
At restart, if the client doesn't register durable interest for exactly the same keys as before then the entries
in the interest list are not copied from the server during the registration. Instead, the client cache starts out
empty and entries are added during updates. If no updates come in for an entry, it never shows up in the
client cache.
Sending Cache Ready Messages to the Server
After a durable client connects and initializes its cache, regions, cache listeners, and any interest
registration, it invokes readyForEvents to indicate to the servers that the client is ready to receive any
messages accumulated for it.
Durable Client Cache Ready Notification (C++)
The following example shows how to call readyForEvents.
//Send ready for event message to server(only for durable
clients).
//Server will send queued events to client after receiving this.
cachePtr->readyForEvents();
To keep the client from losing events, do not call this method until all regions and listeners
are created. For more information, see Reconnection.
151
Preserving Data
Pivotal GemFire Native Client
Disconnecting from the Server
When a durable client closes its cache and disconnects, it tells the servers whether to maintain its queues.
For this purpose, use the version of Cache::close with the boolean keepalive parameter set, as
shown in the following example. If the setting is true, the servers keep the durable client’s queues and
durable subscriptions alive for the timeout period. In addition to in-memory queue retention, the servers
can evict the most recent durable client queue updates to disk to reduce memory consumption.
Only the resources and data related to the session are removed, such as port numbers and non-durable
subscriptions. If the setting is false, the servers do the same cleanup that they would do for a nondurable
client.
Durable Client Disconnect With Queues Maintained
// Close the Cache and disconnect with keepalive=true.
// Server will queue events for durable registrations and CQs
// When the client reconnects (within a timeout period) and
sends
// "readyForEvents()", the server will deliver all stored events
cachePtr->close(true);
Life Cycle of a Durable Client
This section discusses the high-level operation of a durable client through initial startup, disconnection, and
reconnection.
Initial Operation
The initial startup of a durable client is similar to the startup of any other client, except that it specifically
calls the Cache.readyForEvents method when all regions and listeners on the client are ready to
process messages from the server.
See Sending the Cache Ready Message to the Server.
Disconnection
While the client and servers are disconnected, their operation varies depending on the circumstances.
Normal disconnect
When a durable client disconnects normally, the Cache.close request states whether to maintain the
client's message queue and durable subscriptions. The servers stop sending messages to the client and
release its connection. SeeDisconnecting From the Server for more information.
If requested, the servers maintain the queues and durable interest list until the client reconnects or times
out. The non-durable interest list is discarded. The servers continue to queue up incoming messages for
entries on the durable interest list. All messages that were in the queue when the client disconnected
remain in the queue, including messages for entries on the non-durable list.
If the client requests to not have its subscriptions maintained, or if there are no durable subscriptions, the
servers unregister the client and perform the same cleanup as for a non-durable client.
152
Preserving Data
Pivotal GemFire Native Client
Abnormal disconnect
If the client crashes or loses its connections to all servers, the servers automatically maintain its message
queue and durable subscriptions until the client reconnects or times out.
Client disconnected but operational
If the client operates while it is disconnected, it gets what data it can from the local cache. Since updates
are not allowed, the data can become stale. An UnconnectedException occurs if an update is
attempted.
Timing out while disconnected
The servers track how long to keep a durable client queue alive based on the durable-clienttimeout setting. If the client remains disconnected longer than the timeout, the servers unregister the
client and do the same cleanup that is performed for a non-durable client. The servers also log an alert.
When a timed-out client reconnects, the servers treat it as a new client making its initial connection.
Reconnection
During initialization, operations on the client cache can come from multiple sources.
•
•
•
Cache operations by the application.
Results returned by the cache server in response to the client’s interest registrations.
Callbacks triggered by replaying old events from the queue.
These procedures can act on the cache concurrently, and the cache is never blocked from doing
operations.
GemFire handles the conflicts between the application and interest registration, but you need to prevent
the callback problem. Writing callback methods that do cache operations is never recommended, but it is a
particularly bad idea for durable clients, as explained in Implementing Cache Listeners for Durable Clients
on page 161.
Program the durable client to perform these steps, in order, when it reconnects:
1. Create the cache and regions. This ensures that all cache listeners are ready. At this point, the
application hosting the client can begin cache operations.
2. Issue its register interest requests. This allows the client cache to be populated with the initial interest
registration results. The primary server responds with the current state of those entries if they still exist
in the server’s cache.
3. Call Cache.readyForEvents. This tells the servers that all regions and listeners on the client are
now ready to process messages from the servers. The cache ready message triggers the queued
message replay process on the primary server.
For an example that demonstrates Cache.readyForEvents, see Sending the Cache Ready Message to
the Server.
This figure shows the concurrent procedures that occur during the initialization process. The application
begins operations immediately on the client (step 1), while the client's cache ready message (also step 1)
triggers a series of queue operations on the cache servers (starting with step 2 on the primary server). At
the same time, the client registers interest (step 2 on the client) and receives a response from the server.
Message B2 applies to an entry in Region A, so the cache listener handles B2's event. Because B2 comes
before the marker, the client does not apply the update to the cache.
153
Preserving Data
Pivotal GemFire Native Client
Figure 6: Initialization of a Reconnected Durable Client
Only one region is shown for simplicity, but the messages in the queue could apply to multiple regions.
Also, the figure omits the concurrent cache updates on the servers, which would normally be adding more
messages to the client's message queue.
Durable Message Replay
When the primary server receives the cache ready message, the servers and client execute a procedure to
update the queue and replay the events from the stored messages.
Durable message replay proceeds as follows. To avoid overwriting current entries with old data, the client
does not apply the updates to its cache.
1. The server finds the queue for this durable client ID and updates its information, including the client’s
socket and remote ports.
If the client has timed out while it was disconnected, its queues are gone and the server then treats it as
a new client. See Initial Operation on page 157.
2. All servers that have a queue for this client place a marker in the queue.
Messages in the queue before the marker are considered to have come while the client was
disconnected. Messages after the marker are handled normally.
154
Preserving Data
Pivotal GemFire Native Client
3. The cache server sends the queued messages to the client. This includes any messages that were
evicted to disk.
4. The client receives the messages but does not apply the updates to its cache. If cache listeners are
installed, they handle the events. For implications, see Implementing Cache Listeners for Durable
Clients.
5. The client receives the marker message indicating that all past events have been played back.
6. The cache server sends the current list of live regions.
7. In each live region on the client, the marker event triggers the afterRegionLive callback.
After the callback, the client begins normal processing of events from the server and applies the
updates to its cache.
Even when a new client starts up for the first time, the cache ready markers are inserted in the queues.
If messages start coming into the new queues before the servers insert the marker, those messages are
considered as having happened while the client was disconnected, and their events are replayed the same
as in the reconnect case.
Application Operations During Interest Registration
As soon as the client creates its regions, the application hosting the client can start cache operations, even
while the client is still receiving its interest registration responses.
In that case, application operations take precedence over interest registration responses.
When adding register interest responses to the cache, the following rules are applied:
•
•
•
If the entry already exists in the cache with a valid value, it is not updated.
If the entry is invalid and the register interest response is valid, the valid value is put into the cache.
If an entry is marked destroyed, it is not updated. Destroyed entries are removed from the system after
the register interest response is completed.
If the interest response does not contain any results because all of those keys are absent from the server’s
cache, the client’s cache can start out empty. If the queue contains old messages related to those keys,
the events are still replayed in the client’s cache.
Implementing Cache Listeners for Durable Clients
A cache listener for durable clients requires all callback methods to behave properly when stored events
are replayed. A cache listener has a callback method, afterRegionLive, specifically for durable clients
aspects.
For general instructions on implementing a cache listener, see CacheListener.
Writing Callbacks for Use With Durable Messaging
Durable clients require special attention to cache callbacks generated by the cache listener. During the
initialization window when a reconnecting client has a functioning cache but is still receiving the stored
messages from the queue, the client can replay events that are long past. These events are not applied to
the cache, but they are sent to the cache listener. If the listener’s callbacks invoked by these events make
changes to the cache, that could conflict with current operations and create data inconsistencies.
Consequently, you need to keep your callback implementations lightweight and not do anything in the
cache that could produce incorrect results during this window. For details on implementing callbacks for
GemFire event handlers, seeImplementing Cache Event Handlers in the GemFire User’s Guide.
Implementing the afterRegionLive Method
If you are using cache listeners, you can implement the afterRegionLive callback method provided for
durable clients. This callback is invoked when the client has received all the old messages that were stored
155
Preserving Data
Pivotal GemFire Native Client
in its queue while it was disconnected. Implementing this method enables you to do application-specific
operations when the client has replayed all of these old events.
If you do not wish to use this callback, and your listener is an instance of CacheListener (not a
CacheListenerAdapter), you must implement afterRegionLive as a non-operational method.
156
Preserving Data
Pivotal GemFire Native Client
157
Security
Pivotal GemFire Native Client
Part
IX
Security
Topics:
•
•
•
•
•
Authentication
Encrypted Authentication
Client Authorization
Security-Related System
Properties (gfcpp.properties)
SSL Client/Server
Communication
Security describes how to implement the security framework for the
Pivotal GemFire native client, including authentication, authorization,
ecryption, and SSL client/server communication.
Security describes how to implement the security framework for the
Pivotal GemFire native client, including authentication, authorization,
ecryption, and SSL client/server communication.
The security framework authenticates clients that attempt to connect to
a GemFire cache server, and authorizes client cache operations. You
can also configure it for client authentication of servers, and you can
plug in your own implementations for authentication and authorization.
158
Authentication
Pivotal GemFire Native Client
Authentication
A client is authenticated when it connects, with valid credentials, to a GemFire cache server that is
configured with the client Authenticator callback.
A client is authenticated when it connects, with valid credentials, to a GemFire cache server that is
configured with the client Authenticator callback.
Once the client is authenticated, the server assigns the client a unique ID and principal, used to authorize
operations. The client must trust all cache servers in the server system as it may connect to any one of
them. For information on configuring client/server , see Client/Server Configuration in the Topologies and
Communication section of the GemFire User's Guide.
Process and Multiuser Authentication
Client connections can be authenticated at two levels, process and multiuser.
•
•
Process. Each pool creates a configured minimum number of connections across the server group.
The pool accesses the least-loaded server for each cache operation.
Process-level connections represent the overall client process and are the standard way a client
accesses the server cache.
Multi-user. Each user/pool pair creates a connection to one server and then sticks with it for
operations. If the server is unable to respond to a request, the pool selects a new one for the user.
Typically, application servers or web servers that act as clients to GemFire servers make multi-user
connections. Multi-user allows a single application or web server process to service a large number of
users with varied access permissions.
By default, server pools use process-level authentication. Enable multi-user authentication by setting a
pool's multi-user-secure-mode-enabled attribute to true.
159
Security
Pivotal GemFire Native Client
Credentials can be sent in encrypted form using the Diffie-Hellman key exchange algorithm. See Encrypt
Credentials with Diffe-Hellman on page 164 for more information.
Configuring Credentials for Authentication
The native client uses system properties to acquire valid credentials for authentication by the server. You
define these properties in the gfcpp.properties file, which the native client accesses during startup.
•
•
•
•
security-client-auth-factory on page 160
security-client-auth-library on page 160
Implementing the Factory Method for Authentication (C++ and .NET) on page 161
Acquiring Credentials Programmatically (C++ and .NET) on page 161
security-client-auth-factory
System property for the factory function of the class implementing the AuthInitialize interface
(IAuthInitialize in .NET). The .NET clients can load both C++ and .NET implementations. For .NET
implementations, this property is the fully qualified name of the static factory function (including the
namespace and class).
security-client-auth-library
System property for the library where the factory methods reside. The library is loaded explicitly
and the factory functions are invoked dynamically, returning an object of the class implementing the
AuthInitialize interface.
Other implementations of the AuthInitialize interface may be required to build credentials using
properties that are also passed as system properties. These properties also start with the securityprefix. For example, the PKCS implementation requires an alias name and the corresponding keystore
path, which are specified as security-alias and security-keystorepath, respectively.
160
Security
Pivotal GemFire Native Client
Similarly, UserPasswordAuthInit requires a username specified in security-username, and the
corresponding password is specified in the security-password system property.
The getCredentials function for the AuthInitialize interface is called to obtain the credentials.
All system properties starting with security- are passed to this callback as the first argument to the
getCredentials function, as shown in the following code snippet:
PropertiesPtr getCredentials(PropertiesPtr& securityprops, const char
*server);
Implementing the Factory Method for Authentication (C++ and .NET)
The following examples show how to implement the factory method in both C++ and .NET. C++
Implementation
LIBEXP AuthInitialize* createPKCSAuthInitInstance()
{
return new PKCSAuthInit( );
}
.NET Implementation
public static IAuthInitialize Create()
{
return new UserPasswordAuthInit();
}
Implementations of the factory method are user-provided. Credentials in the form of properties returned by
this function are sent by the client to the server for authentication during the client’s handshake process
with the server.
The GemFire native client installation provides sample security implementations in its templates/
security folder.
Acquiring Credentials Programmatically (C++ and .NET)
This example shows a C++ client connecting with credentials.
PropertiesPtr secProp = Properties::create();
secProp->insert("security-client-auth-factory",
"createPKCSAuthInitInstance");
secProp->insert("security-client-auth-library", "securityImpl");
secProp->insert("security-keystorepath", "keystore/gemfire6.keystore");
secProp->insert("security-alias", "gemfire6");
secProp->insert("security-zkeystorepass", "gemfire");
CacheFactoryPtr cacheFactoryPtr = CacheFactory::createCacheFactory(secProp);
This example shows a .NET client.
Properties secProp = Properties.Create();
secProp.Insert("security-client-auth-factory",
"GemStone.GemFire.Templates.Cache.Security.
UserPasswordAuthInit.Create");
secProp.Insert("security-client-auth-library", "securityImpl");
secProp.Insert("security-username"," gemfire6");
secProp.Insert("security-password"," gemfire6Pass);
161
Security
Pivotal GemFire Native Client
Configuring Authentication by the Cache Server
When the cache server receives client credentials during the handshake operation, the server
authenticates the client with the callback configured in the security-client-authenticator system
property. The handshake succeeds or fails depending on the results of the authentication process.
Here is an example of how you could configure security-client-authenticator in the
gfcpp.properties file:
security-client-authenticator=templates.security.PKCSAuthenticator.create
In the preceding configuration sample, PKCSAuthenticator is the callback class implementing the
Authenticator interface and create is its factory method.
The following example shows an implementation of the static create method:
public static Authenticator create() {
return new PKCSAuthenticator();
}
Server Authentication Errors
An AuthenticationRequiredException is thrown when the server is configured with security
and the client does not present its credentials while attempting to connect. This can occur if the
securityclient-auth-factory and security-client-auth-library properties are not
configured on the client.
Creating Multiple Secure User Connections
To create multiple, secure connections to your servers from a single client, so the client can service
different user types, you create an authenticated RegionService for each user.
Typically, a GemFire client embedded in an application server supports data requests from many users.
Each user can be authorized to access a subset of data on the servers. For example, customer users are
allowed only to see and update their own orders and shipments.
The authenticated users all access the same Cache through instances of the RegionService interface.
See RegionService.
To implement multiple user connections in your client cache, create your Cache as usual, with these
additions:
1. Configure your client’s server pool for multiple secure user authentication. Example:
<pool name="serverPool" multiuser-authentication="true">
<locator host="host1" port="44444"/>
</pool>
This enables access through the pool for the RegionService instances and disables it for the Cache
instance.
2. After you create your cache, for each user, call your Cache instance createAuthenticatedView
method, providing the user’s particular credentials. These are create method calls for two users:
PropertiesPtr credentials1 = Properties::create();
credentials1->insert("security-username", "root1");
162
Security
Pivotal GemFire Native Client
credentials1->insert("security-password", "root1");
RegionServicePtr userCache1 =
cachePtr->createAuthenticatedView( credentials1 );
PropertiesPtr credentials2 = Properties::create();
credentials2->insert("security-username", "root2");
credentials2->insert("security-password", "root2");
RegionServicePtr userCache2 =
cachePtr->createAuthenticatedView( credentials2 );
For each user, do all of your caching and region work through the assigned region service pointer. Use
the region service to get your regions, and the query service, if you need that, and then do your work
with them. Access to the server cache will be governed by the server’s configured authorization rules
for each individual user.
3. To close your cache, close the Cache instance.
Requirements and Caveats for RegionService
For each region, you can perform operations through the Cache instance or the RegionService
instances, but not both.
Note: Through the Cache you can create a region that uses a pool configured for multi-user
authentication, then access and do work on the region using your RegionService instances.
To use RegionService:
•
•
Configure regions as EMPTY. Depending on your data access requirements, this configuration might
affect performance, because the client goes to the server for every get.
If you are running durable CQs through the region services, stop and start the offline event storage
for the client as a whole. The server manages one queue for the entire client process, so you need
to request the stop and start of durable client queue (CQ) event messaging for the cache as a whole,
through the ClientCache instance. If you closed the RegionService instances, event processing
would stop, but the events from the server would continue, and would be lost.
Stop with:
cachePtr->close(true);
Start up again in this order:
1. Create the cache.
2. Create all region service instances. Initialize CQ listeners.
3. Call the cache readyForEvents method.
Using an LDAP Server for Client Authentication
An LDAP server can be used by a GemFire cache server using the sample LDAP implementation provided
in GemFire server product.
See the Security chapter in the GemFire User's Guide to verify authentication credentials for native clients
attempting to connect to the GemFire servers and sending username and passwords using the sample
UserPassword scheme.
Note: The username and password with this sample implementation is sent out in plaintext. For
better security, either turn on credential encryption using Diffie-Hellman key exchange, or use a
scheme like PKCS.
When a client initiates a connection to a cache server, the client submits its credentials to the server and
the server submits those credentials to the LDAP server. To be authenticated, the credentials for the client
163
Encrypted Authentication
Pivotal GemFire Native Client
need to match one of the valid entries in the LDAP server. The credentials can consist of the entry name
and the corresponding password. If the submitted credentials result in a connection to the LDAP server
because the credentials match the appropriate LDAP entries, then the client is authenticated and granted a
connection to the server. If the server fails to connect to the LDAP server with the supplied credentials then
an AuthenticationFailedException is sent to the client and its connection with the cache server is closed.
Configuration Settings
In the gfcpp.properties file for the client, specify the UserPasswordAuthInit callback, the
username, and the password, like this:
security-client-auth-library=securityImpl
security-client-auth-factory=createUserPasswordAuthInitInstance
security-username=<username>
security-password=<password>
For server side settings and LDAP server configuration, see the Security chapter in the GemFire User's
Guide.
Encrypted Authentication
You can set up encrypted authentication using Diffe-Hellman or the sample PKCS implementation.
You can set up encrypted authentication using Diffe-Hellman or the sample PKCS implementation.
Encrypt Credentials with Diffe-Hellman
For secure transmission of sensitive credentials like passwords, encrypt credentials using the DiffieHellman key exchange algorithm. With Diffie-Hellman enabled, you can have your client authenticate its
servers.
•
•
Enabling Diffe-Hellman on page 164
Client Authentication of Server on page 165
Enabling Diffe-Hellman
Set the security-client-dhalgo system property in the gfcpp.properties file to the password for
the public key file store on the client (the name of a valid symmetric key cipher supported by the JDK).
Valid security-client-dhalgo property values are DESede, AES, and Blowfish, which enable the
Diffie-Hellman algorithm with the specified cipher to encrypt the credentials.
For the AES and Blowfish algorithms, optionally specify the key size for the security-clientdhalgo property. Valid key size settings for the AES algorithm are AES:128, AES:192, and AES:256.
The colon separates the algorithm name and the key size. For the Blowfish algorithm, key sizes from
128 to 448 bits are supported. For example:
security-client-dhalgo=Blowfish:128
For AES algorithms, you may need Java Cryptography Extension (JCE) Unlimited Strength Jurisdiction
Policy Files from Sun or equivalent for your JDK.
Adding settings for Diffie-Hellman on clients also enables challenge response from server to client in
addition to encryption of credentials using the exchanged key to avoid replay attacks from clients to
servers. Clients can also enable authentication of servers, with challenge-response from client to server to
avoid server-side replay attacks.
164
Security
Pivotal GemFire Native Client
Client Authentication of Server
With Diffie-Hellman enabled, you can have your client authenticate its servers.
1. Generate a .pem file for each pkcs12 keystore:
a. Enter this command from a pkcs12 file or a pkcs keystore:
[email protected]: ~> openssl pkcs12 -nokeys -in <keystore/pkcs12 file> -out
<outputfilename.pem >
b. Concatenate the generated .pem files into a single .pem file. You will use its name in the next step.
2. In the gfcpp.properties file:
a. Set security-client-kspath to the name of the .pem file password for the public key file store
on the client.
b. Set security-client-kspasswd to the password for the public key file store on the client.
Using PKCS for Encrypted Authentication
This section discusses the concepts and configurations for the sample UserPassword and PKCS
implementations. Descriptions of their interfaces, classes, and methods are available in the online API.
Note: Native client samples are provided in source form only in the "templates" directory within the
product directory.
Disclaimer: These security samples serve only as example implementations. The implementation and its
source code is provided on an "as-is" basis, without warranties or conditions of any kind, either express
or implied. You can modify these samples to suit your specific requirements and security providers.
GemStone Systems, Inc. takes no responsibility and accepts no liability for any damage to computer
equipment, companies or personnel that might arise from the use of these samples.
With PKCS, clients send encrypted authentication credentials in the form of standard PKCS signatures
to a GemFire cache server when they connect to the server. The credentials consist of the alias name
and digital signature created using the private key that is retrieved from the provided keystore. The server
uses a corresponding public key to decrypt the credentials. If decryption is successful then the client
is authenticated and it connects to the cache server. For unsuccessful decryption, the server sends an
AuthenticationFailedException to the client, and the client connection to the cache server is
closed.
When clients require authentication to connect to a cache server, they use the PKCSAuthInit class
implementing the AuthInitialize interface to obtain their credentials. For the PKCS sample provided
by GemFire, the credentials consist of an alias and an encrypted byte array. The private key is obtained
from the PKCS#12 keystore file. To accomplish this, PKCSAuthInit gets the alias retrieved from the
security-alias property, and the keystore path from the security-keystorepath property.
PKCSAuthInit also gets the password for the password-protected keystore file from the securitykeystorepass property so the keystore can be opened.
Building the securityImpl Library
To use the PKCS sample implementation, you need to build OpenSSL and then build the securityImpl
library. In the gfcpp.properties file for the client, specify the PKCSAuthInit callback, the keystore
path, the security alias, and the keystore password, like this:
security-client-auth-library=securityImpl
security-client-auth-factory=createPKCSAuthInitInstance
security-keystorepath=<PKCS#12 keystore path>
security-alias=<alias>
security-keystorepass=<keystore password>
165
Client Authorization
Pivotal GemFire Native Client
For server side settings, see the description of PKCS sample in the Security chapter in the GemFire User's
Guide.
Client Authorization
Using a provided callback that implements the AccessControl interface, you can configure each server
to authorize some or all cache operations.
Using a provided callback that implements the AccessControl interface, you can configure each server
to authorize some or all cache operations.
The callback can also modify or even disallow the data being provided by the client in the operation, such
as a put or a putAll operation. The callback can also register itself as a post-processing filter that is
passed operation results like get, getAll, and query.
Configuring Client Authorization
You can configure authorization on a per-client basis for various cache operations such as creates,
gets, puts, query invalidations, interest registration, and region destroys. On the server side, the
securityclient-accessor system property in the server’s gemfire.properties file specifies the
authorization callback.
For example:
security-client-accessor=templates.security.XmlAuthorization.create
In the preceding system property setting, XmlAuthorization is the callback class that implements
the AccessControl interface. The XmlAuthorization sample implementation provided with
GemFire expects an XML file that defines authorization privileges for the clients. For details of this sample
implementation and the AccessControl interface, see the Authorization Example section in the Security
chapter for the GemFire User's Guide.
Post-Operative Authorization
Authorization in the post-operation phase occurs on the server after the operation is complete and before
the results are sent to the client.
The callback can modify the results of certain operations, such as query, get and keySet, or even
completely disallow the operation. For example, a post-operation callback for a query operation can filter
out sensitive data or data that the client should not receive, or even completely fail the operation.
The security-client-accessor-pp system property in the server’s gemfire.properties file
specifies the callback to invoke in the post-operation phase. For example:
security-client-accessor-pp=templates.security.XmlAuthorization.create
Determining Pre- or Post-Operation Authorization
The OperationContext object that is passed to the authorizeOperation method of the callback
as the second argument provides an isPostOperation method that returns true when the callback is
invoked in the post-operation phase.
For example:
bool authorizeOperation(Region region, OperationContext context)
166
Security-Related System Properties (gfcpp.properties)
Pivotal GemFire Native Client
{
If(context.isPostOperation())
{
//It's a post-operation
}
else
{
//it's a pre-operation
}
}
If an authorization failure occurs in a pre-operation or post-operation callback on the server, the operation
throws a NotAuthorizedException on the client.
For more information, see Authorization in the GemFire User’s Guide.
Security-Related System Properties
(gfcpp.properties)
The table describes the security-related system properties in the gfcpp.properties file for native client
authentication and authorization.
The table describes the security-related system properties in the gfcpp.properties file for native client
authentication and authorization.
Table 19: System Properties for Client Authentication and Authorization
security-client-auth-factory
Sets the key for the AuthInitialize factory function.
security-client-auth-library
Registers the path to the securityImpl.dll library.
security-client-dhalgo
Returns the Diffie-Hellman secret key cipher algorithm.
security-client-kspath
Path to a .pem file, which contains the public certificates for
all GemFire cache servers to which the client can connect
through specified endpoints.
security-client-kspasswd
Password for the public key file store on the client.
security-keystorepath
Path to the public keystore.
security-alias
Alias name for the key in the keystore.
security-keystorepass
Sets the password for the password-protected keystore.
ssl-enabled
Enables SSL-based client/server communication when set
to true. When true, the other ssl-* settings are required. The
default is false, which causes communication to use plain
socket connections.
ssl-keystore
Name of the .PEM keystore file, containing the client’s private
key. Not set by default. Required if ssl-enabled is true.
ssl-keystore-password
Sets the password for the private key PEM file for SSL.
ssl-truststore
Name of the .PEM truststore file, containing the servers’ public
certificate. Not set by default. Required if ssl-enabled is
true.
167
SSL Client/Server Communication
Pivotal GemFire Native Client
SSL Client/Server Communication
This section describes how to configure OpenSSL; implement SSL-based communication between your
clients and servers; and run clients and servers with SSL enabled.
Step 1: Download and install OpenSSL
The open-source OpenSSL toolkit provides a full-strength general purpose cryptography library to operate
along with the PKCS sample implementation for encrypted authentication of native client credentials.
Follow these instructions to download and install OpenSSL for your specific operating system.
The GemFire native client requires OpenSSL 1.0.1h or later. For Windows platforms, you can use either
the regular or the OpenSSL "Light" version.
Note: If you use Cygwin, it is recommended that you do not use the OpenSSL library that comes
with Cygwin because it is built with cygwin.dll as a dependency.
Linux
Download the OpenSSL tarball archive from the OpenSSL web site at http://www.openssl.org/source/.
Copy the downloaded tarball file into NativeClient_xxxx/templates/security/openssl/Linux
and run buildit.sh.
Solaris
Download the OpenSSL tarball archive from the OpenSSL web site at http://www.openssl.org/source/.
Copy the downloaded tarball file into NativeClient_xxxx/templates/security/openssl/SunOS
and run buildit.sh.
Windows
Download the installer for OpenSSL 1.0.1h or later from http://www.openssl.org/related/binaries.html. You
can also use the OpenSSL "Light" version.
Use the downloaded OpenSSL installer to install it on Windows. You can usually accept the default
installation path (C:\OpenSSL).
Step 2: Create keystores
The GemFire server requires keys and keystores in the Java Key Store (JKS) format while the native client
requires them in the clear PEM format. Thus you need to be able to generate private/public keypairs in
either format and convert between the two using the keytool utility and the openssl command.
There are public third party free tools and source code available to download such as the "KeyTool IUI"
tool.
168
Security
Pivotal GemFire Native Client
Step 3: Configure environment variables
Configure your system environment to build and run OpenSSL. Follow the environment setup that applies
to your operating system.
For all references to the Pivotal_GemFire_NativeClient_32or64bit_xxxx directory, replace 32or64bit with
the appropriate architecture and xxxx with the actual four-digit product build number.
Bourne and Korn shells (sh, ksh, bash)
% OPENSSL=<parent folder for OpenSSL binaries>; export OPENSSL
% GFCPP=<path to installation, typically C:
\Pivotal_GemFire_NativeClient_32or64bit_xxxx>; export GFCPP
% LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$GFCPP/lib:$GFCPP/ssl_libs:$OPENSSL/lib
% export LD_LIBRARY_PATH
% CLASSPATH=$GEMFIRE/lib/gfSecurityImpl.jar:$CLASSPATH
Windows
> set GFCPP=<path to installation, typically C:
\Pivotal_GemFire_NativeClient_32or64bit_xxxx>
> set OPENSSL=<path to installed OpenSSL, typically C:\OpenSSL>
> set PATH=<path to Java JDK or JRE>\bin;%GFCPP%\bin;%GFCPP%\ssl_libs;
%OPENSSL%\bin;%PATH%
> set CLASSPATH=<path to GemFire installation>\lib\gfSecurityImpl.jar;
%CLASSPATH%
Step 4: Configure SSL properties in gfcpp.properties and
gemfire.properties
Configure SSL properties.
1. In gfcpp.properties, set ssl-enabled to true and set ssl-keystore and ssl-truststore to
point to your keystore files. See Security-Related System Properties (gfcpp.properties) on page 167
for a description of these properties.
2. On each locator, set the following SSL properties in the locator’s gemfire.properties file:
ssl-enabled=true
ssl-protocols=any
ssl-require-authentication=true
ssl-ciphers=SSL_RSA_WITH_NULL_SHA
Step 5: Start and stop the client and server
Before you start and stop the client and server, make sure you configure the native client with the SSL
properties as described and with the servers or locators specified as usual.
Specifically, ensure that:
•
•
OpenSSL and ACE_SSL DLLs locations are in the right environment variables for your system: PATH
for Windows, and LD_LIBRARY_PATH for Unix.
You have generated the keys and keystores.
169
Security
Pivotal GemFire Native Client
•
You have set the system properties.
For details on stopping and starting locators and cache servers with SSL, see Starting Up and Shutting
Down Your System in the GemFire User’s Guide.
Example locator start command
gfsh>start locator --name=my_locator --port=12345 --dir=. \
--J=-Djavax.net.ssl.keyStore=gemfire.jks,\
-Djavax.net.ssl.keyStorePassword=gemfire,\
-Djavax.net.ssl.trustStore=gemfire.jks,\
-Djavax.net.ssl.trustStorePassword=gemfire
These SSL settings can also be configured in your server's gfsecurity.properties file. If you are
using this file, you can start your locator as follows:
gfsh>start locator --name=my_locator --port=12345 --dir=. \
--security-properties-file=/path/to/your/gfsecurity.properties
Example locator stop command
gfsh>stop locator --port=12345 \
--J=-Djavax.net.ssl.keyStore=gemfire.jks,\
-Djavax.net.ssl.keyStorePassword=gemfire,\
-Djavax.net.ssl.trustStore=gemfire.jks,\
-Djavax.net.ssl.trustStorePassword=gemfire
Or if you configured your SSL settings in gfsecurity.properties:
gfsh>stop locator --port=12345 \
--security-properties-file=/path/to/your/gfsecurity.properties
Example server start command
gfsh>start server --name=my_server --locators=hostname[12345] \
--cache-xml-file=server.xml --log-level=fine \
--J=-Xmx1024m,-Xms128m,-Djavax.net.ssl.keyStore=gemfire.jks,\
-Djavax.net.ssl.keyStorePassword=gemfire,\
-Djavax.net.ssl.trustStore=gemfire.jks,\
-Djavax.net.ssl.trustStorePassword=gemfire
Or if you configured your SSL settings in gfsecurity.properties:
gfsh>start server --name=my_server --locators=hostname[12345] \
--cache-xml-file=server.xml --log-level=fine \
--security-properties-file=/path/to/your/gfsecurity.properties
Example server stop command
gfsh>stop server --name=my_server
Limitations
Currently the native client only supports the NULL cipher with mutual authentication for SSL socket
communications.
The keys and keystores need to be in the JKS (Java KeyStore) format for the GemFire server and in the
clear PEM format for the native client.
170
Security
Pivotal GemFire Native Client
171
Remote Querying
Pivotal GemFire Native Client
Part
X
Remote Querying
Topics:
•
•
•
•
•
Remote Querying Basics
Using Query Strings in the
Native Client
Accessing Cached Data
Query Language Elements
Remote Query API
Remote Querying documents remote querying from the native client
to the GemFire cache server. Using examples and procedures, it
describes how to use the APIs to run queries against cached data;
work with query strings in the native client; create and manage
queries; and create indexes.
Remote Querying documents remote querying from the native client
to the GemFire cache server. Using examples and procedures, it
describes how to use the APIs to run queries against cached data;
work with query strings in the native client; create and manage
queries; and create indexes.
172
Remote Querying Basics
Pivotal GemFire Native Client
Remote Querying Basics
Use the Pivotal GemFire native client query API to query your cached data stored on a GemFire cache
server. The query is evaluated and executed on the cache server, and the results are returned to the native
client.
Use the Pivotal GemFire native client query API to query your cached data stored on a GemFire cache
server. The query is evaluated and executed on the cache server, and the results are returned to the native
client.
You can also optimize your queries by defining indexes on the cache server.
The query language for the native client is essentially a subset of OQL (ODMG 3.0 Object Data
Management Group, http://www.odmg.org.), which is based on SQL-92. OQL is a SQL-like language with
extended functionality for querying complex objects, object attributes, and methods.
It is assumed that you have general familiarity with SQL querying and indexing, and with the information on
the native client cache provided in previous chapters of this guide.
The online C++ and .NET API documentation located in the docs directory for the native client provides
extensive details for all of the querying interfaces, classes and methods.
Query language features and grammar are described in detail in the Querying chapter of the GemFire
User's Guide. This chapter describes areas that are unique to the native client.
If you are using the new pool API, you should obtain the QueryService from the pool . For information
about the pool API, see Native Client Pool API on page 217.
Examples Data and Class Definitions
Examples show C++ and corresponding Java class definitions and sample data for the /portfolios
region. The region's keys are the portfolio ID.
User-defined data types must implement the Serializable interface on the native client side, while
corresponding Java classes must implement the DataSerializable interface. The C++ objects for the
native client must correspond to the Java objects for the GemFire cache server. This means that an object
on one side should deserialize correctly to the other side.
Sample C++ class definition
class Portfolio : public Serializable {
int ID;
char * type;
char * status;
Map<Position> positions;
}
class Position : public Serializable {
char * secId;
double mktValue;
double qty;
}
Corresponding Java class definition
class Portfolio implements DataSerializable {
int ID;
173
Remote Querying
Pivotal GemFire Native Client
String type;
String status;
Map positions;
}
class Position implements DataSerializable {
String secId;
double mktValue;
double qty;
}
The following table lists the sample data in the /portfolios region.
Table 20: Entry values in /portfolios
id
type
Statusted
Position:
secID
Position:
mktValue
Position:
qty
111
xyz
active
xxx
27.34
1000.00
xxy
26.31
1200.00
xxz
24.30
1500.00
222
xyz
active
yyy
18.29
5000.00
333
abc
active
aaa
24.30
10.00
333
abc
active
aab
23.10
15.00
444
abc
inactive
bbb
50.41
100.00
444
abc
inactive
bbc
55.00
90.00
Because the client cache waits during transaction execution, and client regions are not distributed, the only
activities that interact with a client transaction are those that occur on the server.
Executing a Query from the Native Client
C#/.NET and C++ examples show how to execute a query from the native client.
Note: In all queries that use the example data, it is assumed that the /portfolios region has
javaobject.Portfolio objects on the cache server.
1. If you are using the C++ native client, get a pointer to the QueryService method.
2. Create a QueryPtr to a query (C++) or create a query instance (C# .NET) that is compatible with the
OQL specification.
3. Use the execute method for the Query interface to submit the query string to the cache server. The
server remotely evaluates the query string and returns the results to the client.
4. You can iterate through the returned objects as part of the query process.
C#/.NET Example
Query<Portfolio> qry = qrySvc.NewQuery("SELECT DISTINCT * FROM
/Portfolios");
ISelectResults<Portfolio> results = qry.Execute();
SelectResultsIterator<Portfolio> iter = results.GetIterator(); while
(iter.MoveNext()) {
Console.WriteLine( iter.Current.ToString()); }
174
Remote Querying
Pivotal GemFire Native Client
C++ Example
Note: The C++ examples in this chapter all assume that you have already obtained a pointer to the
QueryService.
QueryServicePtr qrySvcPtr =
cachePtr->getQueryService("examplePool");
QueryPtr qry = qrySvcPtr->newQuery(
"SELECT DISTINCT * FROM /Portfolios WHERE status = ‘active’");
SelectResultsPtr resultsPtr = qry->execute(10);
SelectResultsIterator iter = resultsPtr->getIterator();
while(iter.hasNext())
{
PortfolioPtr portfolio = dynCast<PortfolioPtr > (iter.next());
}
Querying the Portfolios Region
Examples show a sampling of specific queries that you can run against /portfolios on the server.
The query results for the data are listed in the previous table, Entry values in /portfolios. For the first
several, the coding examples are included as well to show how you can execute the queries using the API.
Get distinct positions from portfolios with at least a $25.00 market value
This query assigns iterator variable names to the collections in the FROM clause. For example, the
variable qryP is the iterator for the entry values in the /portfolios region. This variable is used in the
second part of the FROM clause to access the values of the positions map for each entry value.
Query string:
SELECT DISTINCT posnVal
FROM /portfolios, positions.values posnVal TYPE Position
WHERE posnVal.mktValue >= 25.00
Results:
Collection of Position instances with secId: xxx, xxy, bbb, bbc
Retrieve all active portfolios
In the following example, a query response timeout parameter of 10 seconds is specified for the execute
method to allow sufficient time for the operation to succeed.
Query string:
SELECT DISTINCT * FROM /portfolios WHERE status = ‘active’
Results:
A collection of Portfolio objects for IDs 111 , 222 and 333 .
Code:
QueryServicePtr qrySvcPtr =
cachePtr->getQueryService("examplePool");
QueryPtr qry = qrySvcPtr->newQuery(
"SELECT DISTINCT * FROM /Portfolios WHERE status = ‘active’");
SelectResultsPtr resultsPtr = qry->execute(10);
SelectResultsIterator iter = resultsPtr->getIterator();
while(iter.hasNext())
{
PortfolioPtr portfolio = dynCast<PortfolioPtr >(iter.next());
175
Remote Querying
Pivotal GemFire Native Client
}
Retrieve all active portfolios that have type xyz
The type attribute is passed to the query engine in double quotes to distinguish it from the query keyword
of the same name. A query response timeout parameter of 10 seconds is specified for the execute method
to allow sufficient time for the operation to succeed.
Query string:
SELECT DISTINCT * FROM /portfolios
WHERE status = 'active' AND "type" = 'xyz'
Results:
A collection of Portfolio objects for IDs 111 and 222
Code:
QueryServicePtr qrySvcPtr =
cachePtr->getQueryService("examplePool");
QueryPtr qry = qrySvcPtr->newQuery("SELECT DISTINCT * FROM
/Portfolios WHERE status = 'active' and \"type\"='xyz'");
SelectResultsPtr results = qry->execute(10);
SelectResultsIterator iter = results->getIterator();
while(iter.hasNext())
{
PortfolioPtr portfolio = dynCast<PortfolioPtr >(iter.next());
}
Get the ID and status of all portfolios with positions in secId 'yyy'
Query string:
SELECT DISTINCT id, status FROM /portfolios
WHERE NOT (SELECT DISTINCT * FROM positions.values posnVal TYPE
Position WHERE posnVal.secId='yyy').isEmpty
Results:
A collection of Struct instances, each containing an id field and a status
field.
For this data, the collection length is 1 and the Struct contains data
from the entry with id 222.
Code:
QueryServicePtr qrySrvPtr = cachePtr->getQueryService("examplePool");
QueryPtr qry = qrySvcPtr->newQuery(
"import javaobject.Position; SELECT DISTINCT ID, status FROM "
"/Portfolios WHERE NOT (SELECT DISTINCT * FROM positions.values"
"posnVal TYPE Position WHERE posnVal.secId='DELL').isEmpty");
SelectResultsPtr results = qry->execute(10);
SelectResultsIterator iter = results->getIterator();
while(iter.hasNext())
{
Struct * si = (Struct*) iter.next().ptr();
SerializablePtr id = si->operator[]("ID");
SerializablePtr status = si->operator[]("status");
printf("\nID=%s, status=%s", id->toString()->asChar(), status>toString()->asChar());
}
176
Remote Querying
Pivotal GemFire Native Client
Modifying Cache Contents
To modify the cache based on information retrieved through querying, retrieve the entry keys and use them
in the standard entry update methods.
The query service is a data access tool, so it does not provide any cache update functionality.
The next example shows entry key retrieval.
Get distinct entry keys and positions from active portfolios with at least a $25.00
market value
In the following example, retrieving the entry keys allows you to access the cached region
entries for update. You cannot update the cache through the query engine.
Query string:
SELECT DISTINCT key, posnVal
FROM /portfolios.entrySet, value.positions.values posnVal TYPE
Position
WHERE posnVal.mktValue >= 25.00
Results:
A SelectResults of Struct instances containing key, Position
pairs.
Creating Indexes
An index can provide significant performance gains for query execution. You create and maintain indexes
on the cache server.
A query run without an index iterates through every object in the collection on the cache server. If an index
is available that matches part or all of the query specification, the query iterates only over the indexed set,
and query processing time can be reduced.
When you create your indexes on the cache server, remember that indexes incur maintenance costs as
they must be updated when the indexed data changes. An index that requires many updates and is not
used very often may require more system resources than no index at all. Indexes also consume memory.
For information on the amount of memory used for indexes, see the system configuration information in the
Pivotal GemFire User's Guide.
You can create an index for remote querying declaratively on the cache server in a cache.xml file, as
shown in the next example.
Creating an Index on a Cache Server Using a Server XML File
<region name="portfolios">
<region-attributes . . . >
<value-constraint>cacheRunner.Portfolio</value-constraint>
</region-attributes>
<index name="myFuncIndex">
<functional from-clause="/portfolios" expression="status"/>
</index>
<index name="myPrimIndex">
<primary-key field="id"/>
</index>
177
Remote Querying
Pivotal GemFire Native Client
<entry> . . .
For detailed information about working with indexes configured on a cache server, see the section Working
with Indexes in the GemFire User's Guide.
Remote Querying Requirements
Note the particular requirements for using region endpoints; setting server region data policy and scope;
implementing equals and hashcode methods; and setting object type constraints.
•
•
•
•
Using Region Endpoints on page 178
Setting Server Region Data Policy and Scope on page 178
Implementing the equals and haschcode Methods on page 178
Setting Object Type Constraints on page 179
Using Region Endpoints
When you are using region endpoints, at least one region must exist on the native client before a query can
be executed through the client. All objects in the region belong to the same class hierarchy (homogenous
types).
Setting Server Region Data Policy and Scope
Native client remote querying only accesses the data that is available in the remote cache server region, so
no local cache loading operations are performed. Depending on the cache server region's scope and datapolicy attribute settings, this could mean that your queries and indexes only see a part of the data available
for the server region in the distributed cache.
To ensure a complete data set for your queries and indexes, your cache server region must use one of the
REPLICATE region shortcut settings in the region attribute refid or it must explicitly have its data policy set
to replicate or persistent-replicate .
For a cache server region, setting its data policy to replicate or persistent-replicate ensures that it
reflects the state of the entire distributed region. Without replication, some server cache entries may not be
available.
Depending on your use of the server cache, the non-global distributed scopes distributed-ack and
distributed-no-ack may encounter race conditions during entry distribution that cause the data set
to be out of sync with the distributed region. The global scope guarantees data consistency across the
distributed system, but at the cost of reduced performance.
The following table summarizes the effects of cache server region scope and data policy settings on the
data available to your querying and indexing operations. For more information, see the Distributed and
Replicated Regions in the Pivotal GemFire User's Guide.
Table 21: Effects of Cache Server Region Scope and Data Policy on the Data Available for Querying
Region Scope
Not replicated
Replicated
distributeded-ack or
distributed-no-ack
N/A
FULL data set (if no race conditions).
global
N/A
FULL data set.
Implementing the equals and haschcode Methods
The Portfolio and Position query objects for the cache server must have the equals and hashCode
methods implemented, and those methods must provide the properties and behavior mentioned in the
178
Using Query Strings in the Native Client
Pivotal GemFire Native Client
online documentation for Object.equals and Object.hashCode. Inconsistent query results can occur
if these methods are absent.
See the Object class description in the GemFire online Java API documentation for more information
about the equals and hashCode methods.
Setting Object Type Constraints
Performing queries on cache server regions containing heterogeneous objects, which are objects of
different data types, may produce undesirable results. Queries should be performed only on regions that
contain homogeneous objects of the same object type, although subtypes are allowed.
So your queries will address homogeneous data types, you need to be aware of the values that the client
adds to the server. You can set the key-constraint and value-constraint region attributes to restrict
region entry keys and values to a specific object type. However, because objects put from the client remain
in serialized form in the server cache and do not get deserialized until a query is executed, it is still possible
to put heterogeneous objects from the client.
See the vFabric GemFire User's Guide for descriptions of the key-constraint and valueconstraint attributes for the cache server. See Specifying the object types of FROM clause collections
for more information on associating object types with queries.
Using Query Strings in the Native Client
To use a query string in a native client, specify the string as a parameter in a QueryService::newQuery
method, then execute the query using Query::execute, passing in the required parameters.
To use a query string in a native client, specify the string as a parameter in a QueryService::newQuery
method, then execute the query using Query::execute, passing in the required parameters.
Alternatively, if an expression evaluates to a boolean value, you can specify it using the region shortcut
methods Region::existsValue, Region::selectValue, and Region::query. These shortcut
methods evaluate whether given expressions return any entries and return a single value entry,
respectively. See Region Shortcut Query Methods for more information about these shortcut methods.
If your query requires any IMPORT statements, you must include these before the SELECT statement in
the query string that is passed to the query engine. It should be a fully qualified package name relative to
the cache server. The Java class definition must exist and have the exact footprint as the native client C++
class.
FROM Clause
The FROM clause establishes collections of objects that are iterated over by the remainder of the query.
The attributes of the objects in these collections are added to the name space scope for the remainder of
the FROM clause as well as for the WHERE clause and the SELECT projection list.
Each FROM clause expression must evaluate to a collection. The expression /portfolios.keySet is
valid because it evaluates to a Collection, but /portfolios.name, which evaluates to a String ,
causes an exception to be thrown.
Like the SQL query, which iterates over the tables named in its FROM clause, the OQL query iterates over
the Collections established in its FROM clause.
In the following query, positions.values evaluates to a Collection because positions is a Map,
and the method values on Map returns a Collection.
IMPORT javaobject.Position;
179
Remote Querying
Pivotal GemFire Native Client
SELECT DISTINCT "type"
FROM /portfolios, positions.values posnVal TYPE Position
WHERE posnVal.qty > 1000.00
Every expression in the FROM clause must evaluate to a Collection. For a Map, the values method
returns a Collection.
If positions were a List instead of a Map , this query could be used to retrieve the data:
IMPORT javaobject.Position;
SELECT DISTINCT "type"
FROM /portfolios, positions posnVal TYPE Position
WHERE posnVal.qty >= 1000.00
A List is a Collection, so you can access it directly or through its toArray method.
For each object type accessed in your FROM clause, use the method that returns a Collection for that
object.
Each expression in the FROM clause can be any expression that evaluates to a Collection. An
expression in the FROM clause is typically a path expression that resolves to a region in the cache so that
the values in the region become the collection of objects to filter.
For example, this is a simple SELECT statement that evaluates to a set of all the entry value objects of the
region /portfolios with active status. The collection of entry values provided by the FROM clause is
traversed by the WHERE clause, which accesses each element’s status attribute for comparison.
SELECT DISTINCT * FROM /portfolios WHERE status = 'active'
If the FROM clause has only one expression in it, the result of the clause is the single collection that the
expression evaluates to. If the clause has more than one expression in it, the result is a collection of
structs that contain a member for each of those collection expressions. For example, if the FROM clause
contains three expressions that evaluate to collections C1, C2, and C3, the FROM clause generates a
set of struct(x1, x2, x3) where x1, x2, and x3 represent nested iterations over the collections
specified.
If the collections are independent of each other, this struct represents their cartesian product.
In this query, the FROM clause produces a struct of portfolio and position pairs to be iterated. Each
element in the struct contains the portfolio and one of its contained positions.
IMPORT javaobject.Position;
SELECT DISTINCT "type" FROM /portfolios, positions TYPE Position
WHERE qty > 1000.00
To understand the effects of FROM expressions on query scope, see Drilling Down for Modifying Query
Scope.
Using Iterator Variables
For each collection expressed in the FROM clause, you can associate an explicit variable.The variable is
added to the current scope and becomes the iterator variable bound to the elements of the collection as
they are iterated over. In this example, pflo and posnVal are both explicit iterator variables.
Query Using Explicit Iterator Variables
IMPORT javaobject.Position;
SELECT DISTINCT pflo."type", posnVal.qty
FROM /portfolios pflo, positions.values posnVal TYPE Position
WHERE pflo.status = 'active' and posnVal.mktValue > 25.00
180
Remote Querying
Pivotal GemFire Native Client
Importing and Using Object Classes
To facilitate the specification of type in variable type declarations and in typecasting expressions, a query
string can have IMPORT statements preceding the declarations. By using IMPORT in the query string, the
client can tell the cache server about the class definition of the serialized object that is present in the cache
server region.
The only place you can have a package name in a query is in an import statement. These are valid:
IMPORT com.myFolder.Portfolio;
IMPORT com.myFolder.Portfolio AS MyPortfolio;
The first form of the import statement allows Portfolio to be used as the name of the class,
com.myFolder.Portfolio. The second form provides an alternative class name, MyPortfolio, to be
used. This is useful when a class name is not unique across packages and classes in a single query.
Using Imported Classes
The following example uses imported classes:
IMPORT com.commonFolder.Portfolio;
IMPORT com.myFolder.Portfolio AS MyPortfolio;
SELECT DISTINCT mpflo.status
FROM /portfolios pflo TYPE Portfolio,
/myPortfolios mpflo TYPE MyPortfolio,
WHERE pflo.status = 'active' and mpflo.id = pflo.id
This entire query string must be passed to the query engine, including the IMPORT
statements.
Common type names do not require an IMPORT statement. The following table lists the
types that are defined by the system and the Java types they represent.
Predefined Class Types
The FROM clause establishes collections of objects that are iterated over by the remainder of the query.
The attributes of the objects in these collections are added to the name space scope for the remainder of
the FROM clause as well as for the WHERE clause and the SELECT projection list.
The type specification can be an imported type or any of these predefined types.
Type
Java
C++
.NET
short
short
CacheableInt16
Int16
long
long
CacheableInt64
Int64
int
int
CacheableInt32
Int32
float
float
CacheableFloat
Single
double
double
CacheableDouble
Double
char
char
CacheableWideChar
Char
string
java.lang.String
CacheableString
String
boolean
boolean
CacheableBoolean
Boolean
byte or octet
byte
CacheableByte
Byte
181
Remote Querying
Pivotal GemFire Native Client
date
java.util.Date
CacheableDate
DateTime
time
java.sql.Time
Unsupported
Unsupported
timestamp
java.sql.Timestamp
Unsupported
Unsupported
set<type>
java.util.Set
CacheableHashSet
HashSet<type>
list<type>
java.util.List
CacheableVector
List<type>
array<type>
java.lang.Object[]
CacheableArray
ArrayList<type>
map<type,type> or
dictionary<type,type>
java.lang.Map
CacheableHashMapp
Dictionary<type, type>
or HashTable
Specifying the Object Types of FROM Clause Collections
To resolve implicit attribute names, the query engine must be able to associate each attribute or method
name to a single iterator expression in the FROM clause.
Depending on the complexity of the query, the engine may be able to discover the proper associations on
its own, but providing the specifications described here increases the chances for success.
The server region being queried should contain only homogeneous objects of the same type. See Setting
Object Type Constraints for more information.
The object type information must be available when the query is created. To provide the appropriate
information to the query engine, specify the type for each of your FROM clause collection objects by
importing the object's class before running the query and typing the object inside the query. For the
example region, this query is valid (all of the examples in this chapter assume that this IMPORT statement
is provided):
Query Using IMPORT and TYPE for Object Typing
IMPORT javaobject.Position;
SELECT DISTINCT mktValue
FROM /portfolios, positions.values TYPE Position
WHERE mktValue > 25.00
This entire query string must be passed to the query engine, including the IMPORT statement. Import the
object’s class before running the query and typecast the object inside the query. For the example region,
both of these queries are valid:
Query Using IMPORT and Typecasting for Object Typing
IMPORT javaobject.Position;
SELECT DISTINCT value.mktValue
FROM /portfolios, (map<string,Position>)positions
WHERE value.mktValue > 25.00
IMPORT cacheRunner.Position;
SELECT DISTINCT mktValue
FROM /portfolios, (collection<Position>)positions.values
WHERE mktValue > 25.00
This entire query string must be passed to the query engine, including the IMPORT statement. Use named
iterators in the FROM clause and explicitly prefix the path expression with iterator names.
Query Using Named Iterators for Object Typing
SELECT DISTINCT posnVal
FROM /portfolios pflo, pflo.positions.values posnVal
182
Remote Querying
Pivotal GemFire Native Client
WHERE posnVal.mktValue >= 25.00
The IMPORT statements in these examples assume that $GEMFIRE/examples/dist/classes is in the
CLASSPATH . This is required so the cache server can process IMPORT statements. The class's package
name cannot be used in the FROM clause. The package name must be specified in an IMPORT statement.
There is one exception to these typing guidelines. If one FROM expression lacks explicit typing, the query
engine associates all unresolved attributes with that expression and creates the query. An exception is
thrown if any of these attributes are not found at execution time.
SELECT Projection List
The projections in the SELECT projection list are used to transform the results of the WHERE search
operation.
You specify the projection list either as * or as a comma delimited list of expressions. For *, the interim
results of the WHERE clause are returned from the query. Otherwise, the set of objects in the interim
results are iterated and the projections applied to each of the objects. During the application of the
projection list, the attributes of the objects being traversed are in scope for name resolution.
You can also specify retrieval of the entry keys in your projection list. This allows you to access the
associated cached entries for modification and other purposes. The following example shows how the
Region entry key can be obtained by using the region entries in the FROM clause and using appropriate
projections. This query runs on the /portfolios region, returning a set of struct<key:string,
id:string, secId:string> where key is the key of the region entry, id is an entry ID, and secId is
a secId of a positionsmap for the entry.
SELECT DISTINCT key, entry.value.id, posnVal.secId
FROM /portfolios.entrySet entry, entry.value.positions.values posnVal
WHERE entry.value."type" = 'xyz' AND posnVal.secId = 'XXX'
SELECT Statement Query Results
The result of a SELECT statement is a collection that implements the SelectResults interface or it is
UNDEFINED.
The SelectResults returned from the SELECT statement is either a collection of objects or a Struct
collection containing the objects. (See also the online API documentation for Query.)
Because a SELECT statement returns a result, it can be composed with other expressions like the following
example:
(SELECT DISTINCT * FROM /portfolios WHERE status = 'active').iterator
A collection of objects is returned in two cases:
•
•
When only one expression is specified by the projection list and that expression is not explicitly
specified using the fieldname:expression syntax
When the SELECT list is * and a single collection is specified in the FROM clause
Table 22: Matrix of SelectResults Contents Based on SELECT and FROM Clause Specifications
SELECT
*
Single Expressions
Multiple Expressions
183
Remote Querying
Pivotal GemFire Native Client
FROM
single expression
Objects
multiple expressions
Struct
Objects. (Struct if the
projection specifies a
field name.)
Objects. (Struct if the
projection specifies a
field name.)
Struct
Struct
When a Struct is returned, the name of each field in the Struct is determined as follows:
If a field is specified explicitly using the fieldname:expression syntax, the fieldname is used.
If the SELECT projection list is * and an explicit iterator expression is used in the FROM clause, the
iterator variable name is used as the field name.
If the field is associated with a region or attribute path expression, the last attribute name in the
expression is used.
•
•
•
If names can not be decided based on these rules, arbitrary unique names are generated by the query
processor.
These examples show how the projections and FROM clause expressions are applied.
SELECT <*> FROM <single
expression>
SELECT DISTINCT *
FROM /portfolios
Returns the Collection of
active portfolios objects.
WHERE status ='active'
SELECT <single
expression> FROM
<multiple expression>
(without fieldName mentioned)
IMPORT
javaobject.Position;
SELECT DISTINCT secId
FROM /portfolios,
Returns the Collection of
secIds (CacheableString
objects) from the positions of
active portfolios.
positions.values TYPE Position
WHERE status ='active'
SELECT <single
expression> FROM
IMPORT
Returns struct<secIdField:
javaobject.Position;SELECT CacheableString> for the
DISTINCT
active portfolios. (Compare to the
<multiple expression> (with
secIdFieldName:secId
results for the prior query.)
fieldName mentioned)
FROM /portfolios,
positions.values TYPE
Position
WHERE status ='active'
SELECT <*> FROM <multiple IMPORT
expression>
javaobject.Position;
SELECT DISTINCT *
FROM /portfolios,
positions.values TYPE
Position
Returns a Collection of
struct<portfolios:
Portfolio, values:
Position> for the active
portfolios.
WHERE status = 'active'
184
Remote Querying
SELECT <multiple
expression> FROM
<multiple expression>
Pivotal GemFire Native Client
IMPORT
javaobject.Position;
SELECT DISTINCT pflo,
posn
Returns a Collection of
struct<pflo: Portfolio,
posn: Position> for the active
portfolios.
FROM /portfolios pflo,
positions posn TYPE
Position
WHERE pflo.status =
'active'
WHERE Clause
The optional WHERE clause defines the search criteria for the selection, filtering the set of elements
specified by the FROM clause.
Without a WHERE clause, the SELECT projection list receives the entire collection or set of collections as
specified in the FROM clause.
The query processor searches the collection for elements that match the conditions specified in the
WHERE clause conditions. If there is an index on an expression matched by the WHERE clause, then the
query processor may use the index to optimize the search and avoid iterating over the entire collection. For
more information on indexes, see the Advanced Querying chapter in the GemFire User's Guide.
A WHERE clause expression is a boolean condition that is evaluated for each element in the collection.
If the expression evaluates to true for an element, the query processor passes that element on to the
SELECT projection list. This example uses the WHERE clause to return the portfolio objects in the region
that have a type xyz .
SELECT DISTINCT * FROM /portfolios WHERE "type" = 'xyz'
The next query returns the set of all portfolios with a type of xyz and active status.
SELECT DISTINCT * FROM /portfolios WHERE "type" = 'xyz' AND status =
'active'
Joins
If collections in the FROM clause are not related to each other, you can use the WHERE clause to join them.
The statement below returns all the persons from the /Persons region with the same name as a flower in
the /Flowers region.
SELECT DISTINCT p FROM /Persons p, /Flowers f WHERE p.name = f.name
Indexes are supported for region joins. To create indexes for region joins, you create single-region indexes
for both sides of the join condition. These are used during query execution for the join condition.
185
Accessing Cached Data
Pivotal GemFire Native Client
Accessing Cached Data
Accessing your cached data through the querying service is similar to accessing database contents
through SQL queries. How you specify your regions and region contents is particular to the native client.
Accessing your cached data through the querying service is similar to accessing database contents
through SQL queries. How you specify your regions and region contents is particular to the native client.
The query language supports drilling down into nested object structures. Regions can contain nested data
collections that are unavailable until referenced in the FROM clause.
This discussion describes how to navigate to your cached data through the native client query service.
Note: Querying and indexing only operate on remote cache server contents.
Basic Region Access
n the context of a query, you specify the name of a region by its full path, starting with a forward slash (/ ).
Object Attributes
You can access the Region object's public fields and methods from a region path, referred to as
the region's attributes. Using this method, /portfolios.name returns "portfolios" and /
portfolios.name.length returns 10 . An attribute is mapped to a Java class member in three possible
ways with the following priority until a match is found. If the attribute is named x , then:
public method getX()
public method x()
public field x
Note: The term attribute in this context is not the same as a region attribute.
Region Data
You can also access entry keys and entry data through the region:
•
•
•
•
/portfolios.keySet returns the Set of entry keys in the region
/portfolios.entrySet returns the Set of Region.Entry objects
/portfolios.values returns the Collection of entry values
/portfolios return the Collection of entry values
Note: These collections are immutable. Invoking modifier methods on them, such as add and
remove, result in an UnsupportedOperationException.
For the last two bullets, the FROM clause /portfolios.values and /portfolios return the same
thing.
186
Remote Querying
Pivotal GemFire Native Client
Attribute Visibility
Within the current query scope, you can access any available object or object attribute.
In querying, an object's attribute is any identifier that can be mapped to a public field or method in the
object.
In the FROM specification, any object that is in scope is valid, so at the beginning of a query all cached
regions and their attributes on the cache server are in scope.
This query is valid because name resolves to the Region method getName:
/portfolios.name
This query is valid because toArray resolves to the Collection method with the same name:
SELECT DISTINCT * FROM /portfolios.toArray
You cannot, however, refer to the attribute of a collection object in the region path expression where the
collection itself is specified. The following statement is invalid because neither Collection nor Region
contain an attribute named positions. The entry values collection (specified by /portfolios) that does
contain an attribute named positions is not yet part of the query name space.
/* INCORRECT: positions is not an attribute of Region or of Collection */
SELECT DISTINCT * FROM /portfolios.positions
The following SELECT statement is valid because positions is an element of the entry value collection
that is specified by /portfolios. The entry value collection is in scope as soon as the specification in the
FROM expression is complete (before WHERE or SELECT are evaluated).
SELECT DISTINCT positions FROM /portfolios
You can also refer to positions inside the FROM clause after the /portfolios entry value collection is
created. In this example, positions is an element of the /portfolios entry value collection and values is
an attribute of positions:
IMPORT javaobject.Position;
SELECT DISTINCT posnVal
FROM /portfolios, positions.values posnVal TYPE Position
WHERE posnVal.mktValue >= 25.00
After the comma in the FROM clause, /portfolios is in scope, so its value collection can be iterated. In
this case, this is done with the second FROM clause specification, positions.values.
Modifying Query Scope
The query engine resolves names and path expressions according to the name space that is currently in
scope in the query. This is not the region scope attribute, but the scope of the query statement.
The initial name space for any query is composed of the region paths of the cache on the cache server
and the attributes of those paths. New name spaces are brought into scope based on the FROM clause in
the SELECT statement. For example, in this query the FROM expression evaluates to the collection of entry
values in /portfolios. This is added to the initial scope of the query and status is resolved within the
new scope.
SELECT DISTINCT *
187
Remote Querying
Pivotal GemFire Native Client
FROM /portfolios
WHERE status = 'active'
Each FROM clause expression must resolve to a collection of objects available for iteration in the query
expressions that follow. In the example above, /portfolios resolves to the Collection of entry values
in the region. The entry value collection is iterated by the WHERE clause, comparing the status field to the
string active. When a match is found, the value object is added to the return set.
In the following query, the collection specified in the first FROM clause expression is used by the second
FROM clause expression and by the projections of the SELECT statement.
IMPORT cacheRunner.Position;
SELECT DISTINCT "type"
FROM /portfolios, positions.values posnVal TYPE Position
WHERE posnVal.qty > 1000.00
Note: You cannot change the order of the expressions in this FROM clause. The second
expression depends on the scope created by the first expression.
Nested Query Scopes
You can nest scopes by using nested SELECT statements. Names in an inner scope hide identical names
in an outer scope.
In the query below, the inner SELECT creates a new scope, the positions of the current portfolio, inside
the outer SELECT 's scope, /portfolios. This inner scope (the collection of entry values from the /
portfolios region) is first searched for the secId element. The outer scope is searched only if the
secId element is not found in the inner scope.
IMPORT javaobject.Position;
SELECT DISTINCT * FROM /portfolios
WHERE NOT
(SELECT DISTINCT * FROM positions.values TYPE Position
WHERE secId='YYY').isEmpty
This statement shows the outer scope in bold. The outer scope has all the attributes of a Portfolio in it.
IMPORT javaobject.Position;
SELECT DISTINCT * FROM /portfolios
WHERE NOT
(SELECT DISTINCT * FROM positions.values TYPE Position
WHERE secId='YYY').isEmpty
Now the statement with the inner scope is shown in bold. The inner scope has all the attributes of a
Portfolio in it (inherited from the outer scope), and all the attributes of a Position as well.
IMPORT javaobject.Position;
SELECT DISTINCT * FROM /portfolios
WHERE NOT
(SELECT DISTINCT * FROM positions.values TYPE Position
WHERE secId='YYY).isEmpty
188
Query Language Elements
Pivotal GemFire Native Client
When Names Cannot Be Resolved
When a query is executed and a name or path expression resolves to more than one region name
in the scope, or if the name cannot be resolved at all, the client receives a QueryException. The
QueryException contains the message that is generated for the exception that occurs on the server.
Query Language Elements
This section discusses various aspects and tools of the native client query engine.
This section discusses various aspects and tools of the native client query engine.
Method Invocation
The query language supports method invocation inside query expressions.
The query processor maps attributes in query strings using the attribute rules described in Object
attributes. Methods declared to return void evaluate to null when invoked through the query processor.
If you know that the attribute name maps to a public method that takes no parameters, you can simply
include the method name in the query string as an attribute. For example, emps.isEmpty is equivalent to
emps.isEmpty() . In the following example, the query invokes isEmpty on positions , and returns the set
of all portfolios with no positions.
SELECT DISTINCT * FROM /portfolios WHERE positions.isEmpty
The native client also supports the invocation of public methods with parameters. To invoke methods
with parameters, include the method name as an attribute in the query string and provide the method
arguments between parentheses. You can only use constants in the query strings.
Query Language Literals Supported by Native Client
Query language expressions can contain literals as well as operators and attribute names. The native
client supports many types of literals.
•
•
•
•
•
•
•
•
boolean. Boolean value, either TRUE or FALSE.
integer and long. Type long if it is suffixed with the ASCII letter L. Otherwise, it is of type int.
floating point. Type float if it is suffixed with an ASCII letter F. Otherwise, its type is double, and it
can optionally be suffixed with an ASCII letter D. A double or floating point literal can optionally include
an exponent suffix of E or e, followed by a signed or unsigned number.
string. Delimited by single quotation marks. Embedded single quotation marks are doubled. For
example, the character string 'Hello' evaluates to the value Hello, while the character string 'He
said, ''Hello''' evaluates to He said, 'Hello'. Embedded newlines are kept as part of the
string literal.
char. Type char if it is a string literal prefixed by the keyword CHAR; otherwise, it is of type string.
The CHAR literal for the single quotation mark character is CHAR '''' (four single quotation marks).
date. java.sql.Date object that uses the JDBC format prefixed with the DATE keyword: DATE
yyyy-mm-dd. In the Date , yyyy represents the year, mm represents the month, and dd represents the
day. The year must be represented by four digits; a two-digit shorthand for the year is not allowed.
time. Not supported.
timestamp. Not supported.
189
Remote Querying
•
•
•
Pivotal GemFire Native Client
NIL. Equivalent alternative of NULL.
NULL. Same as null in Java.
UNDEFINED. Special literal that is a valid value for any data type. An UNDEFINED value is the result of
accessing an attribute of a null-valued attribute. If you access an attribute that has an explicit value of
NULL, then it is not UNDEFINED. For example, if a query accesses the attribute address.city and
address is NULL, then the result is UNDEFINED. If the query accesses address, then the result is not
UNDEFINED, it is NULL.
Type Conversions
Java rules within a query string require the query processor to perform implicit conversions and promotions
under certain cases in order to evaluate expressions that contain different types.
The query processor performs binary numeric promotion, method invocation conversion, and temporal type
conversion.
•
•
•
Binary numeric promotion on page 190
Method invocation conversion on page 190
Temporal type conversion on page 190
Binary numeric promotion
Binary numeric promotion widens all operands in a numeric expression to the widest representation used
by any of the operands. In each expression, the query processor applies the following rules in order:
•
•
•
•
If either operand is of type double , the other is converted to double .
If either operand is of type float , the other is converted to float .
If either operand is of type long , the other is converted to long .
Both operands are converted to type int .
The query processor performs binary numeric promotion on the operands of the following operators:
•
•
comparison operators <, <=, >, and >=
equality operators = and <>
This is essentially the same behavior as in Java, except that chars are not considered to be numeric in the
native client query language.
Method invocation conversion
Method invocation conversion in the query language follows the same rules as Java method invocation
conversion, except that the query language uses runtime types instead of compile time types, and handles
null arguments differently than in Java. One aspect of using runtime types is that an argument with a null
value has no typing information, and so can be matched with any type parameter. When a null argument
is used, if the query processor cannot determine the proper method to invoke based on the non-null
arguments, it throws an AmbiguousNameException . For more information on method invocation in query
strings, see Method Invocation.
Temporal type conversion
The temporal types that the query language supports on the cache server include the Java types
java.util.Date and java.sql.Date, which are treated the same and can be freely compared
and used in indexes. When compared with each other, java.util.Date and java.sql.Date
types are all treated as nanosecond quantities. The cache server also supports java.sql.Time and
java.sql.Timestamp; however, these types are not supported for conversion by the native client.
190
Remote Query API
Pivotal GemFire Native Client
Remote Query API
You use the native client querying API to access all the querying functionality discussed in the previous
sections.
You use the native client querying API to access all the querying functionality discussed in the previous
sections.
This chapter gives a general overview of the interfaces and classes that are provided by the Query
package API, and the shortcut methods provided in the Region interface. For complete, current information
on the classes and interfaces discussed here, see the native client online API documentation.
Creating and Managing Queries
You create queries on the cache server by obtaining a QueryService method and manage them through
the resulting Query object. The Region interface has several shortcut query methods.
The newQuery method for the Query interface binds a query string. By invoking the execute method, the
query is submitted to the cache server and returns SelectResults, which is either a ResultSet or a
StructSet.
The QueryService method is the entry point to the query package. It is retrieved from the Cache
instance through Cache::getQueryService. If you are using the Pool API you must obtain the
QueryService from the pools and not from the cache.
Query
A Query is obtained from a QueryService method, which is obtained from the cache. The Query
interface provides methods for managing the compilation and execution of queries, and for retrieving an
existing query string.
You must obtain a Query object for each new query. The following example demonstrates the method
used to obtain a new instance of Query:
QueryPtr newQuery(const char * querystr);
Region Shortcut Query Methods
The Region interface has several shortcut query methods. All take a query predicate which is used in
the WHERE clause of a standard query. See WHERE Clause for more information. Each of the following
examples also set the query response timeout to 10 seconds to allow sufficient time for the operation to
succeed.
•
The query method retrieves a collection of values satisfying the query predicate. This call retrieves
active portfolios, which in the sample data are the portfolios with keys 111, 222, and 333:
SelectResultsPtr
results = regionPtr->query("status 'active' ");
•
The selectValue method retrieves one value object. In this call, you request the portfolio with ID
ABC-1 :
SerializablePtr
port = region->selectValue("ID='ABC-1'");
191
Remote Querying
•
Pivotal GemFire Native Client
The existsValue method returns a boolean indicating if any entry exists that satisfies the predicate.
This call returns false because there is no entry with the indicated type:
bool entryExists
= region->existsValue("'type' = 'QQQ' ");
For more information about these shortcut query methods, see the Region class description in the native
client online API documentation.
Query Result Sets
•
•
•
•
SelectResults. Executes the query on the cache server and returns the results as either a ResultSet or
a StructSet .
SelectResultsIterator. Iterates over the items available in a ResultSet or StructSet.
ResultSet. Obtained after executing a Query, which is obtained from a QueryService that is
obtained from a Cache class.
StructSet. Used when a SELECT statement returns more than one set of results. This is accompanied
by a Struct, which provides the StructSet definition and contains its field values.
Query Code Samples Returning ResultSet
API examples demonstrate methods for returning ResultSet for both built-in and user-fined data types.
Query Returns a ResultSet for a Built-In Data Type
QueryServicePtr qrySvcPtr = cachePtr->getQueryService("examplePool");
QueryPtr query =
qrySvcPtr->newQuery("select distinct pkid from /Portfolios");
//specify 10 seconds for the query timeout period
SelectResultsPtr results = query->execute(10);
if (results == NULLPTR)
{
printf( "\nNo results returned from the server");
}
//obtaining a handle to resultset
ResultSetPtr rs(dynamic_cast<ResultSet*> (results.ptr()));
if (rs == NULLPTR)
{
printf ("\nResultSet is not obtained \n"); return;
}
//iterating through the resultset using row index.
for (int32_t row=0; row < rs->size(); row++)
{
SerializablePtr ser((*rs)[row]);
CacheableStringPtr str(dynamic_cast<CacheableString*> (ser.ptr()));
if (str != NULLPTR)
{
printf("\n string column contains - %s \n", str->asChar() );
}
}
Query Returns a ResultSet for a User-Defined Data Type
QueryServicePtr qrySvcPtr = cachePtr->getQueryService("examplePool");
192
Remote Querying
Pivotal GemFire Native Client
const char * querystring = "select distinct * from /Portfolios";
QueryPtr query = qrySvcPtr->newQuery(querystring);
//specify 10 seconds for the query timeout period
SelectResultsPtr results = query->execute(10);
if (results == NULLPTR)
{
printf( "\nNo results returned from the server");
}
//obtaining a handle to resultset
ResultSetPtr rs(dynamic_cast<ResultSet*> (results.ptr()));
if (rs == NULLPTR)
{
printf ("\nResultSet is not obtained \n"); return;
}
//iterating through the resultset using iterators.
SelectResultsIterator iter = rs->getIterator();
while (iter.hasNext())
{
SerializablePtr ser = iter.next();
PortfolioPtr port(dynamic_cast<Portfolio*> (ser.ptr()));
if (port != NULLPTR)
{
printf("\nPortfolio object is - %s \n", port->toString()->asChar() );
}
} // end of rows
Query Code Samples Returning StructSet
These examples return a StructSet for built-in and user-defined data types, Struct objects, and
collections.
Query Returning a StructSet for a Built-In Data Type
QueryServicePtr qrySvcPtr = cachePtr>getQueryService("examplePool");
const char * querystring =
"SELECT DISTINCT ID, pkid, status, getType FROM /Portfolios";
QueryPtr query = qrySvcPtr->newQuery(querystring);
//specify 10 seconds for the query timeout period
SelectResultsPtr results = query->execute(10);
if (results == NULLPTR)
{
printf( "\nNo results returned from the server");
}
//obtaining a handle to resultset
StructSetPtr ss(dynamic_cast<StructSet*> (results.ptr()));
if (ss == NULLPTR)
{
printf ("\nStructSet is not obtained \n");
return;
}
//iterating through the resultset using indexes.
for ( int32_t row=0; row < ss->size(); row++)
{
Struct * siptr = (Struct*) dynamic_cast<Struct*> ( ((*ss)
[row]).ptr() );
if (siptr == NULL)
{
printf("\nstruct is empty \n");
continue;
193
Remote Querying
Pivotal GemFire Native Client
}
//iterate through fields now
for( int32_t field=0; field < siptr->length(); field++)
{
SerializablePtr fieldptr((*siptr)[field]);
if(fieldptr == NULLPTR )
{
printf("\nnull data received\n");
}
CacheableStringPtr
str(dynamic_cast<CacheableString*>(fieldptr.ptr()));
if (str == NULLPTR)
{
printf("\n field is of some other type \n");
}
else
{
printf("\n Data for %s is %s ", siptr>getFieldName(field), str->asChar() );
}
} //end of columns
} // end of rows
Returning Struct Objects
QueryServicePtr qrySvcPtr = cachePtr>getQueryService("examplePool");
const char * querystring =
"SELECT DISTINCT derivedProjAttrbts, key: p.key FROM "
"/Portfolios.entries p, (SELECT DISTINCT x.ID, myPos.secId
FROM "
"/Portfolios x, x.positions.values AS myPos)
derivedProjAttrbts WHERE "
"p.value.ID = derivedProjAttrbts.ID AND
derivedProjAttrbts.secId = 'IBM'";
QueryPtr query = qrySvcPtr->newQuery(querystring);
//specify 10 seconds for the query timeout period
SelectResultsPtr results = query->execute(10);
if (results == NULLPTR)
{
printf( "\nNo results returned from the server");
}
//obtaining a handle to resultset
StructSetPtr ss(dynamic_cast<StructSet*> (results.ptr()));
if (ss == NULLPTR)
{
printf ("\nStructSet is not obtained \n");
return;
}
//iterating through the resultset using indexes.
for ( int32_t row=0; row < ss->size(); row++)
{
Struct * siptr = (Struct*) dynamic_cast<Struct*> ( ((*ss)
[row]).ptr() );
if (siptr == NULL) { printf("\nstruct is empty \n"); }
//iterate through fields now
for( int32_t field=0; field < siptr->length(); field++) {
SerializablePtr fieldptr((*siptr)[field]);
if(fieldptr == NULLPTR )
194
Remote Querying
Pivotal GemFire Native Client
{
printf("\nnull data received\n");
}
CacheableStringPtr
str(dynamic_cast<CacheableString*>(fieldptr.ptr()));
if (str != NULLPTR) {
printf("\n Data for %s is %s ", siptr>getFieldName(field),
str->asChar() );
}
else
{
StructPtr simpl(dynamic_cast<Struct*>
(fieldptr.ptr()));
if (simpl == NULLPTR)
{
printf("\n field is of some other type \n");
continue;
}
printf( "\n struct received %s \n", siptr>getFieldName(field) );
for( int32_t inner_field=0; inner_field < simpl>length(); inner_field++ )
{
SerializablePtr innerfieldptr((*simpl)
[inner_field]);
if (innerfieldptr == NULLPTR)
{
printf("\nfield of struct is NULL\n");
}
CacheableStringPtr
str(dynamic_cast<CacheableString*>
(innerfieldptr.ptr()));
if (str != NULLPTR)
{
printf("\n Data for %s is %s ",
simpl->getFieldName(inner_field),str>asChar() );
}
else
{
printf("\n some other object type inside
struct\n");
}
}
}
} //end of columns
}//end of rows
Returning Collections
QueryServicePtr qrySvcPtr = cachePtr>getQueryService("examplePool");
const char * querystring = "select distinct ID, names from /
Portfolios";
QueryPtr query = qrySvcPtr->newQuery(querystring);
SelectResultsPtr results = query->execute(10);
if (results == NULLPTR) {
printf( "\nNo results returned from the server");
}
195
Remote Querying
Pivotal GemFire Native Client
//obtaining a handle to resultset
StructSetPtr ss(dynamic_cast<StructSet*> (results.ptr()));
if (ss == NULLPTR) {
printf ("\nStructSet is not obtained \n");
return;
}
//iterating through the resultset using indexes.
for ( int32_t row=0; row < ss->size(); row++)
{
Struct * siptr = dynamic_cast<Struct*> ( ((*ss)
[row]).ptr() );
if (siptr == NULL)
{
printf("\nstruct is empty \n");
continue;
}
//iterate through fields now
for( int32_t field=0; field < siptr->length(); field++)
{
SerializablePtr fieldptr((*siptr)[field]);
if(fieldptr == NULLPTR )
{
printf("\nnull data received\n");
}
CacheableStringPtr
str(dynamic_cast<CacheableString*>(fieldptr.ptr()));
if (str != NULLPTR)
{
printf("\n Data for %s is %s ", siptr>getFieldName(field),
str->asChar() );
}
else
{
CacheableObjectArrayPtr
coa(dynamic_cast<CacheableObjectArray*>(fieldptr.ptr()));
if (coa == NULLPTR)
{
printf("\n field is of some other type\n");
continue;
}
printf( "\n objectArray received %s \n",
siptr->getFieldName(field) );
for(unsigned arrlen=0; arrlen < (uint32_t)coa>length(); arrlen++)
{
printf("\n Data for %s is %s ",siptr>getFieldName(field),
coa->operator[](arrlen)->toString()>asChar());
}
}
} //end of columns
}//end of rows
196
Remote Querying
Pivotal GemFire Native Client
197
Continuous Querying
Pivotal GemFire Native Client
Part
XI
Continuous Querying
Topics:
•
•
•
•
How Continuous Querying
Works
Implementing a Continuous
Query
Managing Continuous Queries
CQ API and Main Features
Continuous Querying describes how to implement continuous querying
in the Pivotal GemFire native client so that C++ and C# .NET clients
can run queries against events in the GemFire cache server region. It
also describes main features and the native client CQ API.
198
How Continuous Querying Works
Pivotal GemFire Native Client
How Continuous Querying Works
C++ and C# .NET clients register interest in events using simple query expressions. Events are sent to
client listeners that you can program to do whatever your application requires.
This section covers the following topics:
•
•
•
Overview of CQ Operations on page 199
Logical Architecture and Data Flow on page 199
State and Life Cycle on page 201
Overview of CQ Operations
You subscribe to server-side events using SQL-type query filtering. The native client sends a query to the
server side for execution and receives the events that satisfy the criteria.
For example, in a region storing stock market trade orders, you can retrieve all orders over a certain price
by running a CQ with a query like this:
SELECT * FROM /tradeOrder t WHERE t.price > 100.00
When the CQ is running, the server sends the client all new events that affect the results of the query. On
the native client side, listeners programmed by you receive and process incoming events. For the example
query on /tradeOrder, you might program a listener to push events to a GUI where higher-priced orders
are displayed. CQ event delivery uses the client/server subscription framework described in Client to
Server Connection Process.
CQs do not update the native client region. This is in contrast to other server-to-client messaging, like
the updates sent to satisfy interest registration and responses to get requests from the client. CQs are
notification tools for the CQ listeners, which can be programmed in any way your application requires.
When a CQ is running against a server region, each entry event is evaluated against the CQ query by the
thread that updates the server cache. If either the old or the new entry value satisfies the query, the thread
puts a CqEvent in the client's queue. The CqEvent contains information from the original cache event,
plus information specific to the CQ's execution. Once received by the client, the CqEvent is passed to the
onEvent method of all CqListeners defined for the CQ.
Logical Architecture and Data Flow
Clients can execute any number of CQs, with each CQ given any number of listeners. This figure shows
the logical architecture of continuous querying.
199
Continuous Querying
Pivotal GemFire Native Client
The next figure shows the typical CQ data flow when entries are updated in the server cache. A description
of the data flow follows, along with a description of CQ state and life cycle.
1. Entry events come to the server's cache from any source: the server or its peers, distribution from
remote sites, or updates from a client.
2. For each event, the server's CQ executor framework checks for a match with the CQs it has running.
200
Implementing a Continuous Query
Pivotal GemFire Native Client
3. If the old or new entry value satisfies a CQ query, a CQ event is sent to the CQ's listeners on the client
side. Each listener for the CQ gets the event. In the preceding figure:
•
•
•
Both new and old prices for entry X satisfy the CQ query, so that event is sent indicating an update
to the query results.
The old price for entry Y satisfied the query, so it was part of the query results. The invalidation of
entry Y means that it does not satisfy the query. Because of this, the event is sent indicating that it is
destroyed in the query results.
The price for the newly created entry Z does not satisfy the query, so no event is sent.
The region operations do not translate directly to specific query operations, and the query operations do
not specifically describe the region events. Instead, each query operation describes how its corresponding
region event affects the query results. For more information, see CqEvent Object.
State and Life Cycle
A CQ has three possible states that can be accessed from the client by calling CqQuery.getState.
•
•
•
STOPPED. The CQ has been created but not yet executed, or it has been explicitly stopped from
executing. The stopped CQ uses system resources. You start or restart the CQ by calling the execute
method on CqQuery.
RUNNING. The CQ is being executed on the server for all events in the region referenced by the query.
Results are sent to all client listeners associated with the CqQuery.
CLOSED. The CQ is closed and is not using system resources. Invoking an execute or stop method on
closed CqQuery throws an exception.
Typical CQ life cycle
1. The client creates the CQ. This sets up everything for running the query and provides the client with a
CqQuery object, but does not execute the CQ. At this point, the query is in a STOPPED state, ready to
be closed or run.
2. The client runs the CQ with an API call to one of the CqQuery execute* methods. This puts the
query into a RUNNING state on the client and on the server.
3. The CQ is closed by a client call to CqQuery.close. This de-allocates all resources in use for the CQ
on the client and server. At this point, the cycle could begin again with the creation of a new CqQuery
instance.
Implementing a Continuous Query
You can specify CQs for any client region.
Here are the high-level steps for implementing a continuous query, with links to more detailed information
in this chapter.
1. Make sure your system is configured properly to run CQs. See Configuring for Continuous Querying.
2. Decide what data to track on the server and write your queries. Use your criteria for tracking data
changes to write your CQ queries. See Writing the Continuous Query.
3. Determine how to handle the CQ events on the client and write your listeners.
Each CQ can have any number of listeners. In addition to your core CQ listeners, you might have
listeners that you use for all CQs to track statistics or other general information. See Writing the CQ
Listener.
4. Write the client code to put the queries and their listeners into named CQ queries and execute the
queries. Make sure you close the queries if your client no longer needs them and when the client exits.
See Running the Continuous Query Code.
5. Write any code you need for monitoring and managing your CQ query. See Managing Continous
Queries.
201
Continuous Querying
Pivotal GemFire Native Client
6. Run your code, monitor and tune as needed.
Configuring Your System for Continuous Querying
The continuous query (CQ) functionality requires standard client/server distributed system and cache
configuration settings.
•
•
•
•
The client region must use a pool with subscription-enabled set to true .
If you want your CQs to be highly available, configure your servers for high availability as described
in Configuring Highly Available Servers in the Pivotal GemFire User's Guide. When your servers are
highly available, CQs are registered on primary and secondary servers, and server failover is performed
without any interruption to CQ messaging. CQ events messaging uses the same queues used for
server-to-client messaging.
Note: When CQ is used with high availability, the overhead for CQs is higher than for the keybased interest list registration. CQs are executed on the primary and all secondary servers, so
they require more overall server processing.
To obtain a list of all durable CQs registered on the server, use the
QueryService.getAllDurableCqsFromServer API.
If you want your CQs to be durable, configure your native clients for durable messaging. When your
clients are durable, you can create durable CQs whose events are maintained during client disconnects
and replayed for the client when it reconnects. The process and data flow particular to durable CQs is
described in Durable Client Messaging.
Writing the Continuous Query
Each CQ uses a query and any number of listeners. The query filters the events on the server and the
listener handles the events that make it through the query filter. With the query and the listener in hand,
you can create and execute your query through the API.
This is the basic syntax for the CQ query:
SELECT * FROM /fullRegionPath [iterator] [WHERE clause]
The CQ query must satisfy the standard Pivotal GemFire native client querying specifications described in
Remote Querying on page 172. It also must satisfy these restrictions:
•
•
•
The FROM clause must contain only a single region specification, with optional iterator variable.
The query must be a SELECT expression only, preceded by zero or more IMPORT statements.
This means the query cannot be a statement like /tradeOrder.name or (SELECT * from /
tradeOrder).size.
The CQ query cannot use:
•
•
•
•
•
Cross region joins
Drill-downs into nested collections
DISTINCT
Projections
Bind parameters
Queries not meeting these constraints generate an UnsupportedOperationException from the
QueryServicenewCq method.
202
Continuous Querying
Pivotal GemFire Native Client
Writing the CQ Listener or CQ Status Listener
The following C++ and C# .NET examples show how you might program a simple CqListener or
CqStatusListener to update a display screen based on the CQ events it receives.
The listener retrieves the queryOperation and entry key and value from the CqEvent , then updates the
screen according to the operation type provided in queryOperation.
CQ events do not change your client cache. They are provided as an event service only. This allows you
to have any collection of CQs without storing large amounts of data in your regions. If you need to persist
information from CQ events, program your listener to store the information where it makes the most sense
for your application.
Be very careful if you choose to update your cache from your CqListener. If your listener updates the
region that is queried in its own CQ, the update may be forwarded to the server. If the update on the server
satisfies the same CQ, it may be returned to the same listener that did the update, which could put your
application into an infinite loop. This same scenario could be played out with multiple regions and multiple
CQs if the listeners are programmed to update each other's regions.
CqListener Implementation (C++)
// CqListener class
class TradeEventListener : public CqListener {
public:
void onEvent(const CqEvent& cqEvent) {
// Operation associated with the query op
CqOperation::CqOperationType queryOperation =
cqEvent.getQueryOperation();
// key and new value from the event
CacheableKeyPtr key = cqEvent.getKey();
TradeOrderPtr tradeOrder =
dynCast<TradeOrderPtr>(cqEvent.getNewValue());
if (queryOperation==CqOperation::OP_TYPE_UPDATE) {
// update data on the screen for the trade order
. . .
}
else if (queryOperation==CqOperation::OP_TYPE_CREATE) {
// add the trade order to the screen
. . .
}
else if (queryOperation==CqOperation::OP_TYPE_DESTROY) {
// remove the trade order from the screen
. . .
}
}
void onError(const CqEvent& cqEvent) {
// handle the error
}
void close() {
// close the output screen for the trades
. . .
}
}
CqListener Implementation (C# .NET)
// CqListener class
public class TradeEventListener : ICqListener {
203
Continuous Querying
Pivotal GemFire Native Client
public void onEvent(CqEvent cqEvent) {
// Operation associated with the query op
CqOperationType queryOperation =
cqEvent.getQueryOperation();
// key and new value from the event
ICacheableKey key = cqEvent.getKey();
CacheableString keyStr = key as CacheableString;
IGFSerializable val = cqEvent.getNewValue();
TradeOrder tradeOrder = val as TradeOrder;
if (queryOperation==CqOperationType.OP_TYPE_UPDATE) {
// update data on the screen for the trade order
// . . .
}
else if (queryOperation== CqOperationType.OP_TYPE_CREATE) {
// add the trade order to the screen
// . . .
}
else if (queryOperation== CqOperationType.OP_TYPE_DESTROY)
{
// remove the trade order from the screen
// . . .
}
}
public void onError(CqEvent cqEvent) {
// handle the error
}
// From CacheCallback
public void close() {
// close the output screen for the trades
// . . .
}
}
Writing a CqStatusListener
If you need your CQs to detect whether they are connected to any of the servers that host its subscription
queues, implement a CqStatusListener instead of a CqListener.
CqStatusListener extends the current CqListener, allowing a client to detect when a CQ is
connected and/or disconnected from the server(s). The onCqConnected() method will be invoked
when the CQ is connected, and when the CQ has been reconnected after being disconnected. The
onCqDisconnected() method will be invoked when the CQ is no longer connected to any servers.
Taking the examples from above, we can instead implement a CqStatusListener.
When you install the CqStatusListener, your listener will be able to detect its connection status to the
servers that it is querying.
CqStatusListener Implementation (C++)
class MyCqStatusListener : public CqStatusListener {
uint8_t m_id;
uint32_t m_numInserts;
uint32_t m_numUpdates;
uint32_t m_numDeletes;
uint32_t m_numEvents;
uint32_t m_cqsConnectedCount;
uint32_t m_cqsDisconnectedCount;
public:
uint8_t getId()
204
Continuous Querying
Pivotal GemFire Native Client
{
return m_id;
}
uint32_t getNumInserts()
{
return m_numInserts;
}
uint32_t getNumUpdates()
{
return m_numUpdates;
}
uint32_t getNumDeletes()
{
return m_numDeletes;
}
uint32_t getNumEvents()
{
return m_numEvents;
}
uint32_t getCqsConnectedCount()
{
return m_cqsConnectedCount;
}
uint32_t getCqsDisConnectedCount()
{
return m_cqsDisconnectedCount;
}
MyCqStatusListener(uint8_t id):
m_id(id),
m_numInserts(0),
m_numUpdates(0),
m_numDeletes(0),
m_numEvents(0),
m_cqsDisconnectedCount(0),
m_cqsConnectedCount(0)
{
}
inline void updateCount(const CqEvent& cqEvent)
{
m_numEvents++;
switch (cqEvent.getQueryOperation())
{
case CqOperation::OP_TYPE_CREATE: {
m_numInserts++;
break;
}
case CqOperation::OP_TYPE_UPDATE: {
m_numUpdates++;
break;
}
case CqOperation::OP_TYPE_DESTROY: {
m_numDeletes++;
break;
}
default:
break;
}
}
void onEvent(const CqEvent& cqe){
updateCount(cqe);
}
void onError(const CqEvent& cqe){
updateCount(cqe);
}
205
Continuous Querying
Pivotal GemFire Native Client
void close(){
}
void onCqDisconnected() {
//This is called when the cq loses connection with all
servers.
m_cqsDisconnectedCount++;
}
void onCqConnected() {
//This is called when the cq establishes a connection with a
server.
m_cqsConnectedCount++;
}
void clear() {
m_numInserts = 0;
m_numUpdates = 0;
m_numDeletes = 0;
m_numEvents = 0;
m_cqsDisconnectedCount = 0;
m_cqsConnectedCount = 0;
}
};
CQStatusListener Implementation (C# .NET)
public class MyCqStatusListener<TKey, TResult> :
ICqStatusListener<TKey, TResult>
{
#region Private members
private bool m_failedOver = false;
private UInt32 m_eventCountBefore = 0;
private UInt32 m_errorCountBefore = 0;
private UInt32 m_eventCountAfter = 0;
private UInt32 m_errorCountAfter = 0;
private UInt32 m_CqConnectedCount = 0;
private UInt32 m_CqDisConnectedCount = 0;
#endregion
#region Public accessors
public MyCqStatusListener(int id)
{
}
public void failedOver()
{
m_failedOver = true;
}
public UInt32 getEventCountBefore()
{
return m_eventCountBefore;
}
public UInt32 getErrorCountBefore()
{
return m_errorCountBefore;
}
public UInt32 getEventCountAfter()
{
return m_eventCountAfter;
}
public UInt32 getErrorCountAfter()
{
return m_errorCountAfter;
206
Continuous Querying
Pivotal GemFire Native Client
}
public UInt32 getCqConnectedCount()
{
return m_CqConnectedCount;
}
public UInt32 getCqDisConnectedCount()
{
return m_CqDisConnectedCount;
}
#endregion
}
public virtual void OnEvent(CqEvent<TKey, TResult> ev)
{
if (m_failedOver == true)
m_eventCountAfter++;
else
m_eventCountBefore++;
}
public virtual void OnError(CqEvent<TKey, TResult> ev)
{
if (m_failedOver == true)
m_errorCountAfter++;
else
m_errorCountBefore++;
}
public virtual void Close()
{
}
public virtual void OnCqConnected()
{
m_CqConnectedCount++;
}
public virtual void OnCqDisconnected()
{
m_CqDisConnectedCount++;
}
public virtual void Clear()
{
m_eventCountBefore = 0;
m_errorCountBefore = 0;
m_eventCountAfter = 0;
m_errorCountAfter = 0;
m_CqConnectedCount = 0;
m_CqDisConnectedCount = 0;
}
CqEvent Object
The CqEvent object contains information about the CQ event.
•
•
•
•
Entry key and new value.
Base operation that triggered the CQ event in the server.
CqQuery object associated with this CQ event.
Query operation associated with this CQ event. This operation describes the change affected to the
query results by the cache event. Possible values are:
•
•
CREATE, which corresponds to the standard database INSERT operation.
UPDATE
207
Continuous Querying
•
Pivotal GemFire Native Client
DESTROY, which corresponds to the standard database DELETE operation.
This table describes the query operation based on whether the old and new entry values in the region entry
event satisfy the query criteria.
Table 23: Query Operation Based on Old and New Entry Values
Old Entry Value
New Entry Value
No value or value does not satisfy the No value (operation is
query criteria.
invalidate or destroy) or
value does not satisfy the query.
Query Operation
N/A - no event
create
Value satisfies the query.
Value satisfies the query
No value (operation is
invalidate or destroy) or
value does not satisfy the query.
destroy
update
Value satisfies the query.
You can use the query operation to decide what to do with the CqEvent in your listeners. For example,
a CqListener that displays query results on screen might stop displaying the entry, start displaying the
entry, or update the entry display depending on the query operation.
Running the Continuous Query Code
Create your CQ from an instance of the QueryService . Once created, the CQ is maintained primarily
through the CqQuery interface. The following two C++ and C# examples show the basic calls in the CQ life
cycle.
Create your CQ from an instance of the QueryService. Once created, the CQ is maintained primarily
through the CqQuery interface. The following two C++ and C# examples show the basic calls in the CQ life
cycle.
CQ Creation, Execution, and Close (C++)
// Get cache and qrySvcPtr - refs to local cache and
QueryService
// Create client /tradeOrder region configured to talk to the
server
// Create CqAttribute using CqAttributeFactory
CqAttributesFactory cqf;
// Create a listener and add it to the CQ attributes
// callback defined below
CqListenerPtr tradeEventListener (new TradeEventListener());
QueryServicePtr qrySvcPtr = cachePtr->getQueryService();"
cqf.addCqListener(tradeEventListener);
CqAttributesPtr cqa = cqf.create();
// Name of the CQ and its query
char* cqName = "priceTracker";
char* queryStr = "SELECT * FROM /tradeOrder t where t.price >
100.00";
// Create the CqQuery
CqQueryPtr priceTracker = qrySvcPtr->newCq(cqName, queryStr,
cqa); try {
// Execute CQ
priceTracker->execute();
208
Continuous Querying
Pivotal GemFire Native Client
} catch (Exception& ex){
ex.printStackTrace();
}
// Now the CQ is running on the server, sending CqEvents to
the listener
. . .
}
// End of life for the CQ - clear up resources by closing
priceTracker->close()
CQ Creation, Execution, and Close (C# .NET)
// Get cache and queryService - refs to local cache and
QueryService
// Create client /tradeOrder region configured to talk to the
server
// Create CqAttribute using CqAttributeFactory
CqAttributesFactory cqf = new CqAttributesFactory();
// Create a listener and add it to the CQ attributes
//callback defined below
ICqListener tradeEventListener = new TradeEventListener();
cqf.addCqListener(tradeEventListener);
CqAttributes cqa = cqf.create();
// Name of the CQ and its query
String cqName = "priceTracker ";
String queryStr = "SELECT * FROM /tradeOrder t where t.price
>100.00 ";
QueryService queryService = cache.GetQueryService();
// Create the CqQuery
CqQuery priceTracker = queryService.newCq(cqName, queryStr, cqa,
true);
try {
// Execute CQ
priceTracker.execute();
}catch (Exception ex){
//handle exception
}
// Now the CQ is running on the server, sending CqEvents to the
listener
// . . .
}
// End of life for the CQ - clear up resources by closing
priceTracker.close();
CQ Execution Options
CQ execution can be done with or without an initial result set by calling CqQuery.Execute
or CqQuery.ExecuteWithInitialResults. The initial SelectResults returned from
ExecuteWithInitialResults is the same as the one you would get if you ran the query ad hoc by
calling QueryService.NewQuery.Execute on the server cache, but with the key added.
Individual CQs are executed using CqQueryexecute* methods. You can also execute all CQs for the
client or for a region through the client QueryService. CQs that are running can be stopped or closed.
If you are managing a data set from the CQ results, you can initialize the set by iterating over the result set
and then updating it from your listeners as events arrive. For example, you might populate a new screen
with initial results and then update the screen from a listener.
209
Managing Continuous Queries
Pivotal GemFire Native Client
Just as with the standalone query, the initial results represents a possibly moving snapshot of the cache.
If there are updates to the server region while the result set is being created, the result set and the
subsequent event-by-event CQ query execution might miss some events.
When an Error Occurs in a Running CQ
When an error occurs in CQ execution on the server, specific information on the error itself is stored in the
server's log file. An exception is passed to the client, and the client throws an exception.
The server log will contain an error message indicating an error while processing CQ , like this:
[error 2007/12/18 12:03:18.903 PST gemfire1 <RMI TCP Connection(2)10.80.10.91> tid=0x18] Error while processing CQ on the event, key :
key-1, CqName :testCQEvents_0, ClientId :
identity(carlos(3249):52623/35391,connection=1,durableAttributes=null)
Error :No public attribute named 'ID' was found in class java.lang.Object
Errors in CQ execution are usually caused by data errors, such as invalid object types that are stored in
the server region. In this case, the query is trying to read into an object of type Portfolio for an entry where
an empty object has been stored. You can avoid these types of errors by placing constraints on the region
entries, or by otherwise controlling the types of objects stored in your server regions.
Managing Continuous Queries
This section discusses how to access and manage your CQs from your client. The calls discussed here
are all executed specifically for the calling client. A client cannot access or modify the CQs belonging to
another client.
For detailed method usage, see the online API documentation.
Accessing CQs and CQ Statistics
You can use the QueryServicegetCq* methods to access a single named CQ, an array of all CQs
registered, and an array of all CQs registered in the client. You canuse the CqEvent.getCq method to
access the CQ used to produce a CqEvent.
CQ runtime statistics are available for the client through the CqServiceStatistics and
CqStatistics interfaces described under CQ API and Main Features on page 211. You
can get information on the events generated by a specific CQ from the CqStatistics object
returned by CqQuery.GetStatistics. You can get higher-level information about the CQs the
client has registered, running, and so on, from the CqServiceStatistics object returned by
QueryService.GetCqStatistics .
For both the client and server, you can access these statistics by loading the statistics archive file into
VSD. The VSD (Visual Statistics Display) tool can be acquired as part of the GemFire main product
download.
Client statistics are for the single client only. The server's pertain to all clients with CQs on this server.
Modifying CQ Attributes
You can modify the attributes for an existing CQ using the methods provided by
CqQuery.GetCqAttributesMutator. The attributes consist of a list of listeners.
210
CQ API and Main Features
Pivotal GemFire Native Client
Stopping and Closing CQs
You stop individual CQs with the CqQuerystop method. You can stop all CQs for the client through the
QueryService. Stopped CQs are in the same state as new CQs that have not yet been executed. You
can close or execute a stopped CQ.
You close individual CQs with the CqQueryclose method. You can also close all CQs for the client
through the QueryService. Closed CQs cannot be executed. CQs are also closed in the following cases:
•
•
•
The client closes its cache after closing all of its CQs--Closing the client cache closes the
QueryService and all associated CQs on the client and server.
The client disconnects from its server--This might be because of a network outage or some other
failure. When a client disconnects, all CQs created by the client are removed from the server and put
into a CLOSED state on the client.
The server region is destroyed--When a server region is destroyed, all associated CQs are also
cleaned up on the server and the region destroy event is sent to the client. On the client, the
CqListener.Close method is called for all CQs on the region.
Getting All Durable CQs Registered with the Server
To obtain a list of all durable CQs registered on the server, use the
QueryService.getAllDurableCqsFromServer API.
CQ API and Main Features
The Pivotal GemFire native client API allows your clients to create and manage CQs. (The server
side does not have an API.) Continuous querying provides native client query syntax, events-based
management, integration with client/server architecture, active query execution, and interest criteria based
on data values.
This chapter documents the primary native client API for CQ management. For complete information on
the classes and interfaces described here, see the online API documentation.
•
•
Gemstone::GemFire::Cache on page 211
Main Features of Continuous Querying on page 212
Gemstone::GemFire::Cache
Only C# versions of CQ API interfaces, classes, and methods are shown here (example:
CqQuery.execute). The code examples demonstrate both C++ and C# versions.
•
QueryService interface. Provides methods that enable you to:
•
•
•
•
•
•
•
Create a new CQ and specify whether it is durable (available for durable clients)
Execute a CQ with or without an initial result
List all the CQs registered by the client
Close and stop CQs
Get a handle on CqStatistics for the client
You run QueryService CQ methods against the server cache. The QueryService can be obtained
from the cache or from a pool.
CqQueryinterface. Provides methods for managing a continuous query after it is created through the
QueryService. This interface is used typically to begin and end CQ execution and to retrieve other
CQ-related objects such as CQ attributes, CQ statistics, and CQ state.
CqListener application plug-in interface. Handles continuous query events after they occur. You
program this interface.
211
Continuous Querying
•
•
•
•
Pivotal GemFire Native Client
CqEventinterface. Provides all information sent from the server about the CQ event, which is passed to
the CQ's CqListener methods.
CqState interface. Provides information on the state of a CqQuery, through the getState method of the
CqQuery instance.
CqAttributes, CqAttributesFactory, CqAttributesMutator interfaces. Allow you to
manage CQ attributes. The attributes consist of CqListener plug-in specifications.
CqStatistics, CqServiceStatistics interfaces. Allow you to access statistics information for
a single CQ and for the query service's management of CQs as a whole. For details on statistics, see
Statistics API.
Main Features of Continuous Querying
Continuous querying in the native client has the following features:
•
•
•
•
•
Standard GemFire native client query syntax and semantics. CQ queries are expressed in the
same language used for other native client queries (see Remote Querying.).
Standard GemFire events-based management of CQ events. The event handling used to process
CQ events is based on the standard GemFire Enterprise event handling framework. The CQListener
interface is similar to Application Plug-Ins and Application Callbacks.
Complete integration with the client/server architecture. CQ functionality uses existing server-toclient messaging mechanisms to send events. All tuning of your server-to-client messaging also tunes
the messaging of your CQ events. If your system is configured for high availability then your CQs are
highly available, with seamless failover provided in case of server failure (see High Availability for
Client-to-Server Communication). If your clients are durable, you can also define any of your CQs as
durable (see Durable Client Messaging).
Interest criteria based on data values. CQ queries are run against the region's entry values. Compare
this to register interest by reviewing Registering Interest for Entries.
Active query execution. Once initialized, the queries only operate on new events instead of on the
entire region data set. Events that change the query result are sent to the client immediately.
212
Continuous Querying
Pivotal GemFire Native Client
213
Using Connection Pools
Pivotal GemFire Native Client
Part
XII
Using Connection Pools
Topics:
•
•
How Client Load Balancing
Works
Configuring Pools
Using Connection Pools describes how connection pools achieve load
balancing for the client and describes how to configure connection
pools as server locators or as a list of servers.
214
How Client Load Balancing Works
Pivotal GemFire Native Client
How Client Load Balancing Works
In a distributed system, servers can be added or removed and their capacity to service new client
connections may vary. The server connectivity options are specified in the connection pool configuration.
The Pivotal GemFire connection pool API supports connecting to servers through server locators or directly
connecting to servers.
Server Locators
Server locators continuously monitor server availability and server load information. The client is configured
with a list of server locators and consults a server locator to request a connection to a server in the
distributed system.
Locators provide clients with dynamic server discovery and server load balancing. They give clients
connection information for the server with the least load at any given time.
Server locators provide these main features:
•
•
•
Automated discovery of servers and locators. Adding and removing servers or locators is made
easy as each client does not require a list of servers to be configured at the time of pool creation.
Client load rebalancing. Server locators give clients dynamic server information and provide server
load rebalancing after servers depart or join the system.
High availability. When a client/server connection receives an exception, the connection is
automatically failed over to another available connection in the pool. Redundancy is also provided for
client subscriptions.
Alternatively, you can configure a pool statically with a list of endpoints. When the pools are statically
configured, a round-robin load balancing policy is used to distribute connections across the servers.
Connection Pools
Clients contain connection pools. Each region is associated with a connection pool using a region attribute,
and operations on the region use connections from the respective pools.
The server connectivity options are specified in the connection pool configuration. Each pool has a
minimum and maximum number of connections.
Each cache operation that requires server connectivity obtains a connection from the pool for the server
group that the operation affects, performs the operation using the connection, and returns the connection
to the pool. If the pool size is less than the maximum number of connections and all connections are in
use, the connection pool creates a new connection and returns it. If the pool is at the maximum number of
connections, that thread blocks until a connection becomes available or a free-connection-timeout
occurs. If a free-connection-timeout occurs, an AllConnectionsInUse exception is thrown.
The connection pool has a configurable timeout period that is used to expire idle connections. The idle
connections are expired until the pool has the minimum number of connections. A monitoring thread
expires idle connections, adds sufficient connections to bring up the count to minimum, closes connections
whose lifetime has been exceeded. See the load-conditioning-interval and idle-timeout
attributes of the <pool> element. A separate thread (ping) tests each connected endpoint for its status and
if the endpoint is not reachable, the thread closes all connections that have been made to the endpoint.
See the ping-interval attribute of the <pool> element>.
215
Using Connection Pools
Pivotal GemFire Native Client
When a connection receives an exception, the operation is failed over to another connection from the pool.
The failover mechanism obtains the endpoint to failover to from the locator or from the specified endpoint
list in the pool.
Figure 7: Logical Architecture of Client/Server Connections
Discovering Locators Dynamically
A background thread periodically queries the locator for any other locators joining the distributed system.
To balance load across all the locators, locators are shuffled before pushing in the list.
However, if locator A (to which the client is connected) goes down before it discovers locator B, the locator
B is never discovered even though it is alive and the client receives a NoLocatorsAvailable exception.
One connection is attached to every application thread that is local to the respective thread. This is
known as a thread local connection.
In this case, to perform any cache operation the client is not required to obtain a connection from pool.
Instead the thread local connection of the client is used.
A thread local connection can be released by invoking the
Pool::releaseThreadLocalConnection() method. The released connection is returned to the
pool. If the number of threads is larger than the number of max-connections, the client throws an
AllConnectionsInUseException after the free-connection-timeout lapses, unless the
Pool::releaseThreadLocalConnection() method is used judiciously.
If a connection expires or the server goes down on which the connection was established, a thread local
connection is immediately replaced with a good connection obtained from the pool.
216
Configuring Pools
Pivotal GemFire Native Client
Configuring Pools
A pool can be configured as locators or as a list of servers.
You configure locator, server, and pool settings declaratively in the client's cache.xml file or
programmatically through the PoolFactory method. You create an instance of PoolFactory through
PoolManager.
Native Client Pool API
The GemFire native client API allows your clients to create and manage connection pools. The server side
does not have an API.
This section lists the primary native client API for pool management. For complete information on the
classes and interfaces described here, see the online API documentation.
Note: Only C# versions of Pool API interfaces, classes, and methods are shown throughout the
text in this chapter (example: Pool.GetQueryService()) . The code examples demonstrate
both C++ and C# versions.
Gemstone::GemFire::Cache
•
•
•
•
Pool interface. API to retrieve pool attributes.
PoolFactory interface. API to configure pool attributes.
PoolManager interface. API to create a PoolFactory object and to find the pool objects.
AttributesFactory class. Has a new method setPoolname which assigns a pool to a region.
Operations performed on the configured region use connections from the pool.
Note: A region can have a pool attached to it. A pool may have multiple regions attached to it.
Pool Configuration Example and Settings
Connection pools require standard client/server distributed system and cache configuration settings. You
must also configure settings for the locator, server, and pool elements.
Configure the following settings for the locator, server, and pool elements:
•
•
•
Locator. Host and port where a server locator is listening.
Server. Host and port where a server is listening.
Pool. Client/server connection pool.
The example shows a declarative pool configuration. Following the example is a table that describes the
attributes that can be configured.
Example: Declarative Pool Configuration
This example shows a declarative pool configuration.
Note: You create an instance of PoolFactory through PoolManager.
<pool free-connection-timeout="12345" idle-timeout="5555"
load-conditioning-interval="23456"
217
Using Connection Pools
Pivotal GemFire Native Client
max-connections="7" min-connections="3"
name="test_pool_1" ping-interval="12345"
read-timeout="23456" retry-attempts="3" servergroup="ServerGroup1"
socket-buffer-size="32768" statistic-interval="10123"
subscription-ack-interval="567" subscriptionenabled="true"
subscription-message-tracking-timeout="900123"
subscription-redundancy="0" thread-localconnections="true">
<locator host="localhost" port="34756"/>
</pool>
Pool Attributes
Attribute Name
free-connection-timeout
idle-timeout
load-conditioninginterval
max-connections
Description
Default
Number of milliseconds (ms)
that the client waits for a
free connection if maxconnections limit is configured
and all connections are in use.
10000 ms
Number of milliseconds to wait for 5000 ms
a connection to become idle for
load balancing
Interval in which the pool checks 300000 ms (5 minutes)
to see if a connection to a specific
server should be moved to a
different server to improve the
load balance.
Maximum number of connections
that the pool can create. If all
connections are in use, an
operation requiring a client-to
server-connection is blocked
until a connection is available
or the free-connectiontimeout is reached. If set to -1,
there is no maximum. The setting
must indicate a cap greater than
min-connections.
-1
Note: If you use this
setting to cap your pool
connections, disable
the pool attribute prsingle-hop-enabled.
Leaving single hop
enabled can increase
thrashing and lower
performance.
218
Using Connection Pools
Pivotal GemFire Native Client
Attribute Name
Description
Default
min-connections
Number of connections that must
be created initially.
5
name
Pool name.
ping-interval
Interval between pinging the
server to show the client is
alive, set in milliseconds. Pings
are only sent when the pinginterval elapses between
normal client messages. This
must be set lower than the
server’s maximum-timebetween-pings.
10000 ms
pr-single-hop-enabled
Setting used for singlehop access to partitioned
region data in the servers for
some data operations. See
PartitionResolver. See note in
thread-local-connections
below.
True
read-timeout
Number of milliseconds to wait for 10000
a response from a server before
the connection times out.
retry-attempts
Number of times to retry an
operation after a time-out or
exception for high availability.
If set to -1, the pool tries every
available server once until it
succeeds or has tried all servers.
server-group
Server group from which to select empty
connections. If not specified, the
global group of all connected
servers is used.
socket-buffer-size
Size of the socket buffer, in bytes, 32768
on each connection established.
statistic-interval
Default frequency, in
milliseconds, with which the
client statistics are sent to the
server. A value of -1 indicates
that the statistics are not sent to
the server.
subscription-ack-interval Number of milliseconds
to wait before sending an
acknowledgment to the server
-1
-1
100
219
Using Connection Pools
Attribute Name
Pivotal GemFire Native Client
Description
Default
about events received from the
subscriptions.
subscription-enabled
Whether to establish a server to
client subscription.
False
subscription-messagetracking-timeout
Number of milliseconds for which
messages sent from a server to
a client are tracked. The tracking
is done to minimize duplicate
events.
90000
subscription-redundancy
Redundancy for servers that
contain subscriptions established
by the client. A value of -1
causes all available servers in
the specified group to be made
redundant.
0
thread-local-connections
Whether the connections must
False
have affinity to the thread that last
used them.
Note: To set this to
true, also set prsingle-hop-enabled
to false. A true value
in pr-single-hopenabled automatically
assigns a false value
to thread-localconnections...
Subscription Properties
Each connection pool has a single subscription connection that can be to any server that matches the
requirements of the connection pool.
When a client registers interest for a region, if the connection pool does not already have a subscription
channel, the connection pool sends a message to the server locator, and the server locator chooses
servers to host the queue and return those server names to the client. The client then contacts the chosen
servers and asks them to create the queue.
The client maintains at least one connection with each server hosting a queue. If the server does not detect
any connections from a non-durable client, it drops the client queue and closes all artifacts for the client.
For information about durable client subscriptions, see Durable Client Messaging.
Requesting a Subscription Region Queue
The client-to-server locator request is a short lived TCP request. The client sends a message with:
•
•
•
The client ID.
(Optional) target server group.
Number of redundant copies.
220
Using Connection Pools
•
Pivotal GemFire Native Client
Servers to exclude from the results. This list is used if the client cannot connect to a server and needs
to request a new one.
The server locator responds with a list of servers. The client is responsible for contacting the primary and
secondaries and asking them to host the queue.
For durable subscriptions, the server locator must be able to locate the servers that host the queues for the
durable client. When a durable client sends a request, the server locator queries all the available servers
to see if they are hosting the subscription region queue for the durable client. If the server is located, the
client is connected to the server hosting the subscription region queue.
Running the Connection Pool Code
Examples demonstrate a simple procedure to create a pool factory and then create a pool instance in C++
and C#. They also help you to execute a query.
The examples create a pool with locators. Ensure that you create a pool with locators or endpoints,
but not both. The first example demonstrates creating a pool by adding locators. The second example
demonstrates creating a pool by adding servers. For more information, see the example in the QuickStart
Guide.
Connection Pool Creation and Execution Using C++
PropertiesPtr prptr = Properties::create();
systemPtr = CacheFactory::createCacheFactory(prptr);
cachePtr = systemPtr->create();
PoolFactoryPtr poolFacPtr = PoolManager::createFactory();
//to create pool add either endpoints or add locators or servers
//pool with endpoint, adding to pool factory
//poolFacPtr->addServer("localhost", 12345 /*port number*/);
//pool with locator, adding to pool factory
poolFacPtr->addLocator("localhost", 34756 /*port number*/);
PoolPtr pptr = NULLPTR;
if ((PoolManager::find("examplePool")) == NULLPTR) {
// Pool does not exist with the same name.
pptr = poolFacPtr->create("examplePool");
}
RegionFactoryPtr regionFactory =
cachePtr->createRegionFactory(CACHING_PROXY);
regionPtr = regionFactory
->setPoolName("examplePool")
->create("regionName");
QueryServicePtr qs = cachePtr->getQueryService("examplePool");
Connection Pool Creation and Execution Using (C# .NET)
Properties prop = Properties.Create();
CacheFactory cacheFactory =
CacheFactory.CreateCacheFactory(prop);
Cache cache = cacheFactory.Create();
PoolFactory poolFact = PoolManager.CreateFactory();
//to create pool add either endpoints or add locators
//pool with endpoint, adding to pool factory.
poolFact.AddServer("localhost", 40404 /*port number*/);
//pool with locator, adding to pool factory
//poolFact.AddLocator("hostname", 15000 /*port number*/);
Pool pool = null;
221
Using Connection Pools
Pivotal GemFire Native Client
if (PoolManager.Find("poolName") == null) {
pool = poolFact.Create("poolName");
}
int loadConditInterval = pool.LoadConditioningInterval;
RegionFactory regionFactory =
cache.CreateRegionFactory(RegionShortcut.CACHING_PROXY);
IRegion<string, string> region =
regionFactory.SetPoolName(poolName).Create<string,
string>(regionName);
222
Using Connection Pools
Pivotal GemFire Native Client
223
Transactions
Pivotal GemFire Native Client
Part
XIII
Transactions
Topics:
•
•
•
How Native Client Transactions
Work
Running a Native Client
Transaction
Suspending and Resuming
Transactions
Transactions describes how transactions work on the client side.
It provides examples for running, suspending, and resuming
transactions.
Transactions describes how transactions work on the client side.
It provides examples for running, suspending, and resuming
transactions.
Note: Pivotal GemFire client transactions run on the server
tier. The client uses a server delegate that runs the transaction
as it would a local transaction. Thus the key to running
client transactions lies in making sure the server is properly
configured and programmed. For complete information
about GemFire transactions in the Java server, see the
Transactions chapter in the GemFire User’s Guide. The User’s
Guide provides detailed information including server data
requirements, interactions of GemFire transactions with other
operations running on the server tier, server-side application
plug-ins with transactions, and querying with transactions.
224
How Native Client Transactions Work
Pivotal GemFire Native Client
How Native Client Transactions Work
The syntax for writing client transactions is the same as with server or peer transactions, but when a client
performs a transaction, the transaction is delegated to a server that brokers the transaction.
The syntax for writing client transactions is the same for writing server and peer transactions, but when a
client performs a transaction, the transaction is delegated to a server that brokers the transaction.
•
•
Role of Server Delegates in Transactions on page 225
Native Client Transaction APIs on page 226
Role of Server Delegates in Transactions
The Pivotal GemFire native client can run transactions on the Java cache server, using a server delegate
to actually run the transaction code.
For information on transaction requirements and activities on the server side, see the Transactions chapter
in the GemFire User’s Guide.
Note: The client cache blocks until the transaction is successfully committed. However, the client
cache block is removed if the transaction is suspended.
Depending on where the data resides, the server transaction delegate may or not be the same member
that hosts the transaction. This is the same as for transactions run by the servers, but for server-run
transactions, there is no delegate. There is just the member that is directly running its own transaction
code.
In this figure, the application code on the client makes changes to data entries Y and Z within a transaction.
The server delegate that performs the transaction, M1, does not host the primary copy of the data being
modified. The transaction takes place on server M2, where the data resides.
225
Running a Native Client Transaction
Pivotal GemFire Native Client
To maintain cache consistency, the local client cache is not accessible during a transaction as it may
reflect information inconsistent with the transaction in progress. When the transaction completes, the local
cache is accessible again.
In addition to the failure conditions common to all transactions, client transactions can also fail if the
transaction delegate fails. If the delegate performing the transaction fails, the transaction code throws a
TransactionException.
Figure 8: Transaction Run From a Client
Native Client Transaction APIs
The GemFire API for distributed transactions has the familiar relational database methods, begin,
commit, and rollback. There are also APIs available to suspend and resume transactions.
The .NET classes for executing transactions are:
GemStone::GemFire::Cache::Generic
•
•
CacheTransactionManager
TransactionId
Running a Native Client Transaction
Before you can run a native client transaction, you must configure your clients and servers; define your
server regions for your transactions; and define your client regions.
Before you can run a native client transaction, you must configure your clients and servers; define your
server regions for your transactions; and define your client regions.
For information on server and server region configuration, see the GemFire User’s Guide.
1. Retrieve the GemFire cache transaction manager.
C++ example
CacheTransactionManagerPtr txManager =
cache->getCacheTransactionManager();
C# .NET example
CacheTransactionManager txManager =
cache.CacheTransactionManager;
2. Run your transaction. (Detailed steps follow the examples.)
C++ example
TransactionIdPtr tid;
txManager->begin();
// ..do work
tid = txManager->suspend();
// following code can be run from another
// thread that has access to tid
try {
txManager->resume(tid);
// ..do work
tid = txManager->commit();
catch (const CommitConflictException& e)
}
226
Suspending and Resuming Transactions
Pivotal GemFire Native Client
C# .NET example
TransactionId tid;
txManager.Begin();
// ..do work
tid = txManager.Suspend();
// following code can be run from another
// thread that has access to tid
try {
txManager.Resume(tid);
// ..do work
txManager.Commit();
}
catch (CommitConflictException e)
•
•
•
•
Start each transaction with a begin operation.
If your transaction runs on server regions that are a mix of partitioned and replicated regions,
perform your first transaction operation on a partitioned region. This sets the server data host for
the entire transaction. If you are using PR single-hop, single-hop will be applied as usual to this first
operation.
Run the GemFire operations and other operations that you want included in the transaction.
End each transaction with a commit or a rollback.
Note: Do not leave any transaction in an uncommitted and unrolled back state unless you
have suspended the transaction. Transactions that have not been explicitly suspended do not
time out, so will remain in the system for the life of your application.
3. Review all of your client code for compatibility with transactions.
When you commit a transaction, while the commit is taking place, the changes are visible in the cache.
This is also known as transition commits. This provides better performance than locking everything to do
the transaction updates, but it means that another process accessing data used in the transaction might
get some data in the pre-transaction state and some in the post-transaction state.
For example, key 1 and 2 are written to in a transaction so both of their values change from A to B. In
another thread, it is possible to read key 1 with value B and key 2 with value A, while the transaction is
being committed. This can happen because of how GemFire performs reads. This choice sacrifices atomic
visibility in favor of performance. Reads don't block writes. Writes don't block reads.
Because the client cache waits during transaction execution, and client regions are not distributed, the only
activities that interact with a client transaction are those that occur on the server.
Suspending and Resuming Transactions
The ability to suspend and resume transactions is useful when a thread must perform operations that
should not be part of the transaction before the transaction can complete.
The ability to suspend and resume is useful when a thread must perform some operations that should not
be part of the transaction before the transaction can complete.
When a transaction is suspended, it loses the transactional view of the cache. None of the previous
operations (before calling suspend) are visible to the thread. Subsequently any operations that are
performed by the thread do not participate in the suspended transaction.
When a transaction is resumed, the resuming thread assumes the transactional view. A transaction that is
suspending on a member must be resumed on the same member. Before resuming a transaction, you may
want to check if the transaction exists on the member and whether it is suspended. You may optionally use
the tryResume method.
227
Transactions
Pivotal GemFire Native Client
If the member with the primary copy of the data crashes, the transactional view that applied to that data is
lost. The secondary member for the data cannot resume transactions suspended on the crashed member.
You need to take remedial steps to retry the transaction on a new primary copy of the data.
If a suspended transaction is not touched for a period of time, Pivotal GemFire cleans it up automatically.
By default, the timeout for a suspended transaction is 30 minutes and can be configured by using the
suspended-tx-timeout property of the gfcpp.properties file. The suspended transaction timeout
value is specified in milliseconds.
See Running a Native Client Transaction for code examples of how to suspend and resume a transaction.
228
Transactions
Pivotal GemFire Native Client
229
Function Execution
Pivotal GemFire Native Client
Part
XIV
Function Execution
Topics:
•
•
•
Understanding Data-Aware
Function Routing
How Functions Execute
Executing Functions in GemFire
Function Execution describes how you can execute application
functions to achieve linear scalability. It explains how function
execution works and lists specific use cases.
Note: Function execution can be used only along with
the pool functionality. For more information about the pool
API, see Using Connection Pools. Only C++ versions of
Function Execution API interfaces, classes, and methods (like
FunctionService::onRegion) are shown in text. The code
examples show C++ and C#.
230
Understanding Data-Aware Function Routing
Pivotal GemFire Native Client
Understanding Data-Aware Function Routing
Achieving linear scalability is predicated upon being able to horizontally partition the application data such
that concurrent operations by distributed applications can be done independently across partitions.
Achieving linear scalability is predicated upon being able to horizontally partition the application data such
that concurrent operations by distributed applications can be done independently across partitions.
In other words, if the application requirements for transactions can be restricted to a single partition, and
all data required for the transaction can be colocated to a single server member or a small subset of server
members, then true parallelism can be achieved by vectoring the concurrent accessors to the ever-growing
number of partitions.
Most scalable enterprise applications grow in data volume, where the number of data items managed
rather than the size of individual items grows over time. If the above logic holds (especially true for OLTP
class applications), then we can derive sizable benefits by routing the data-dependent application code to
the fabric member hosting the data. This routing of application code to the data of interest is called dataaware function routing, or behavior routing.
How Functions Execute
This chapter discusses the basic function execution process, how highly available functions execute after a
failure, and the execution scenarios for data-dependent and data-independent functions.
This chapter discusses the basic function execution process, how highly available functions execute after a
failure, and the execution scenarios for data-dependent and data-independent functions.
How Functions Execute
1. The calling client application runs the execute method on the Function Execution object. The
object must be already registered on the servers.
2. The execution is sent to all servers where it needs to run. The locations are determined by the
FunctionService on* method calls, region configuration, and any filters.
3. If the function has results, the result is returned to the execute method call in a ResultCollector
object.
4. The client collects results using the result collector getResult.
How Highly Available Functions Execute after a Failure
If a failure occurs in function execution, the error is returned to the calling application. You can code
for high availability for onRegion functions that return a result, so the function is automatically retried.
For information on setting this up on the server side, see Executing a Function in Pivotal GemFire in
the GemFire User’s Guide. To use a highly available function, the client must call the results collector
getResult method. When an execution error occurs or a member crashes while executing, the system
does the following:
1.
2.
3.
4.
Waits for all calls to return.
Sets a boolean indicating a reexecution is being done.
Calls the result collector’s clearResults method.
Executes the function.
The system retries the execution up to the number specified in the server pool’s retryAttempts setting.
If the function continues to fail, the final exception is returned to the getResult method.
231
Function Execution
Pivotal GemFire Native Client
Data-Independent Function Execution
The figure shows the sequence of events for a data-independent function executed against all available
servers.
Figure 9: Data-Independent Function Invoked from a Client
Data-Dependent Function Execution
The figure shows a data-dependent function run by a client. The specified region is connected to the server
system, so the function automatically goes there to run against all servers holding data for the region.
232
Function Execution
Pivotal GemFire Native Client
Figure 10: Data-Dependent Function Invoked From a Client
This shows the same data-dependent function with the added specification of a set of keys on which to run.
Servers that don't hold any of the keys are left out of the function execution.
233
Function Execution
Pivotal GemFire Native Client
Figure 11: Data-Dependent Function with Filter Invoked from a Client
This scenario demonstrates the steps in a call to a highly available function. The call fails the first time on
one of the participating servers and is successfully run a second time on all servers.
234
Function Execution
Pivotal GemFire Native Client
Figure 12: Highly Available Data-Dependent Function with Failure on First Executions
235
Executing Functions in GemFire
Pivotal GemFire Native Client
Executing Functions in GemFire
Using the Pivotal GemFire function execution service, you can execute application functions on a single
server member, in parallel on a subset of server members, or in parallel on all server members of a
distributed system.
In these procedures, it is assumed that you have defined your client and server regions, and that you have
coded and configured your servers to run your functions. In the Function Execution chapter of the GemFire
User’s Guide, see the function execution information for the server side.
In these procedures, it is assumed that you have defined your client and server regions, and that you
have coded and configured your servers to run your functions. In the "Function Execution" chapter of the
GemFire User’s Guide, see the function execution information for the server side.
Running the Function
In this section you create an Execution object and use its methods to define and run the function. To run
a function with high availability, you call getResult from the results collector returned from the execute
method.
•
•
•
•
•
Configuring and Running a Function on page 236
Running a Function on a Region (C++) on page 237
Running a Function on a Server Pool (C++) on page 237
Running a Function on a Region (C# .NET) on page 237
Running a Function on a Server Pool (C# .NET) on page 238
Configuring and Running a Function
You specify the members that run the function and, optionally, the data set over which the functions run.
•
•
Servers. Execute the function in a single server or a set of servers, specified by the server pool. To
specify data sets for this type of function, pass arguments in to the function.
Data set. Specify a region and possibly a set of keys on which to run.
In every client where you want to execute the function and process the results:
1. Use one of the FunctionService on* methods to create an Execution object. The on* methods,
onRegion, onServer and onServers, define the highest level where the function is run. If you
use onRegion you can further narrow your run scope by setting key filters. The function run using
onRegion is a data dependent function – the others are data-independent functions.
You can run a data dependent function against custom partitioned and colocated partitioned regions.
The steps for setting up the regions and functions in the server are provided in the GemFire User’s
Guide. From the client, just provide the appropriate key sets to the function call.
2. Use the Execution object as needed for additional function configuration. You can:
Provide a set of data keys to withFilter to narrow the execution scope. This works only for
onRegion Execution objects.
• Provide function arguments to withArgs.
• Define a custom ResultCollector to withCollector. See Programming to Get Function
Results on page 238.
3. Call the Execution object execute method to run the function.
4. To run a function with high availability, call getResult from the results collector returned from
execute. Calling a highly available function without using getResult disables the high availability
functionality.
•
236
Function Execution
Pivotal GemFire Native Client
Running a Function on a Region (C++)
regPtr0 = initRegion();
ExecutionPtr exc = FunctionService::onRegion(regPtr0);
CacheableVectorPtr routingObj = CacheableVector::create();
char buf[128];
bool getResult = true;
sprintf(buf, "VALUE--%d", 10);
CacheablePtr value(CacheableString::create(buf));
sprintf(buf, "KEY--%d", 100);
CacheableKeyPtr key = CacheableKey::create(buf);
regPtr0->put(key, value);
sprintf(buf, "KEY--%d", 100);
CacheableKeyPtr key1 = CacheableKey::create(buf);
routingObj->push_back(key1);
CacheablePtr args = routingObj;
CacheableVectorPtr executeFunctionResult = exc>withFilter(routingObj)->
withArgs(args)->execute(func, getResult)->getResult();
Running a Function on a Server Pool (C++)
pptr = PoolManager::find(poolName);
ExecutionPtr exc = FunctionService::onServer(cache);
CacheableVectorPtr routingObj = CacheableVector::create();
char buf[128];
bool getResult = true;
sprintf(buf, "VALUE--%d", 10);
CacheablePtr value(CacheableString::create(buf));
sprintf(buf, "KEY--%d", 100);
CacheableKeyPtr key = CacheableKey::create(buf);
regPtr0->put(key, value);
sprintf(buf, "KEY--%d", 100);
CacheableKeyPtr key1 = CacheableKey::create(buf);
routingObj->push_back(key1);
CacheablePtr args = routingObj;
CacheableVectorPtr executeFunctionResult =
exc->withArgs(args)->execute(func, getResult)->getResult();
Running a Function on a Region (C# .NET)
IRegion<string, string> fregion =
regionFactory.Create<string, string>("exampleRegion");
for (int i = 0; i < 34; i++)
{
fregion.Put("KEY--" + i, "VALUE--" + i, null);
}
object[] routingObj = new object[17];
int j = 0;
for (int i = 0; i < 34; i++)
{
237
Function Execution
Pivotal GemFire Native Client
if (i % 2 == 0) continue;
routingObj[j] = "KEY--" + i;
j++;
}
object args0 = true;
Boolean getResult = true;
// data dependent function execution -- get function with result
Execution<object> exc =
Generic.FunctionService.OnRegion<string, string,
object>(fregion);
Generic.IResultCollector rc =
exc.WithArgs((IGFSerializable)args0).WithFilter(
(IGFSerializable[])routingObj).Execute(getFuncName,
getResult);
object[] executeFunctionResult = rc.GetResult();
Running a Function on a Server Pool (C# .NET)
exc = Generic.FunctionService.OnServer<object>(cache);
List<object> args1 = new List<object>();
for (int i = 0; i < routingObj.Length; i++)
{
Console.WriteLine("routingObj[{0}]={1}.", i, (routingObj[i] as
string));
args1.Add(routingObj[i]);
}
rc = exc.WithArgs((IGFSerializable)args1).Execute(getFuncIName,
getResult);
executeFunctionResult = rc.GetResult();
Console.WriteLine("on one server: result count= {0}.",
executeFunctionResult.Length);
Programming to Get Function Results
GemFire provides a default result collector. If you need special results handling, code a custom
ResultsCollector implementation to replace the default provided by GemFire. Use the
Execution::withCollector method to define your custom collector.
Note: This section applies only to functions that return results.
To program your client to get the results from a function, use the result collector returned from the function
execution, like this:
ResultCollectorPtr rc = FunctionService::onRegion(region)
->withArgs(args)
->withFilter(keySet)
->withCollector(new MyCustomResultCollector())
.execute(Function);
CacheableVectorPtr functionResult = rc.getResult();
The getResult methods of the Pivotal GemFire default result collector block until all results are received,
then return the full result set.
You can handle the results in a custom manner if you wish. To do this:
238
Function Execution
Pivotal GemFire Native Client
1. Write a class that extends ResultCollector and code the methods to handle the results as you
need. The methods are of two types: one handles data and information from GemFire and populates
the results set, while the other returns the compiled results to the calling application:
a. addResult is called by GemFire when results arrive from the Function methods. Use
addResult to add a single result to the ResultCollector.
b. endResults is called by GemFire to signal the end of all results from the function execution.
c. getResult is available to your executing application (the one that calls Execution.execute) to
retrieve the results. This may block until all results are available.
d. clearResults is called by GemFire to clear partial results from the results collector. This is used
only for highly available onRegion functions where the calling application waits for the results. If the
call fails, before GemFire retries the execution, it calls clearResults to ready the instance for a
clean set of results.
2. Use the Execution object in your executing member to call withCollector, passing your custom
collector, as shown in the example above.
Solutions and Use Cases
The function execution service provides solutions for various application use cases.
•
•
•
•
•
An application that executes a server-side transaction or makes data updates using the GemFire
distributed locking service.
An application that initializes some of its components once on each server, which might be used later
by executed functions.
Initialization and startup of a third-party service, such as a messaging service.
Any arbitrary aggregation operation that requires iteration over local data sets that can be done more
efficiently through a single call to the cache server.
Any kind of external resource provisioning that can be done by executing a function on a server.
239
Delta Propagation
Pivotal GemFire Native Client
Part
XV
Delta Propagation
Topics:
•
•
•
•
•
•
How Delta Propagation Works
Delta Propagation API
Cloning
Implementing Delta Propagation
Exceptions and Limitations
Examples of Delta Propagation
Delta Propagation describes how deltas (updates to data) are
propagated and how to implement delta propagation. It also analyzes
performance limitations.
Delta Propagation describes how deltas (updates to data) are
propagated and how to implement delta propagation through clientside methods. It also analyzes performance limitations in some
application scenarios.
In most distributed data management systems, stored data is created
once and updated frequently. Updates are sent to other members for
event propagation, redundancy management, and cache consistency
in general. Tracking only the changes in an updated object and
sending only the updates, or deltas, mean lower network transmission
costs and lower object serialization/deserialization costs. Generally,
the larger your objects and the smaller the deltas, the greater the
performance benefits of delta propagation. Partitioned regions
generally benefit more with higher redundancy levels.
240
How Delta Propagation Works
Pivotal GemFire Native Client
How Delta Propagation Works
Pivotal GemFire propagates object deltas using methods that you program on the client side. The methods
are in the delta interface, which you implement in your cached objects’ classes.
Pivotal GemFire propagates object deltas using methods that you program on the client side. The methods
are in the delta interface, which you implement in your cached objects’ classes.
This figure shows delta propagation for a change to an entry with key, k, and value object, v.
Figure 13: Delta Propagation
1. get operation. The get works as usual; the cache returns the full entry object from the local cache or,
if it is unavailable there, from a server cache or from a loader.
2. update methods. You need to add code to the object’s update methods so that they save delta
information for object updates, in addition to the work they were already doing.
3. put operation. The put works as usual in the local cache, using the full value, then calls hasDelta to
check for deltas and toDelta to serialize the information.
4. receipt of delta. fromDelta extracts the delta information that was serialized by toDelta and
applies it to the object in the local cache. The delta is applied directly to the existing value or to a clone,
depending on how you configure it for the region.
5. additional distributions. As with full distributions, receiving members forward the delta according
to their configurations and connections to other members. In the example, the server would forward
the delta to its peers and its other clients as needed. Receiving members do not recreate the delta;
toDelta is only called in the originating member.
241
Delta Propagation API
Pivotal GemFire Native Client
Delta Propagation API
Delta propagation uses configuration properties and a simple API to send and receive deltas.
Delta propagation uses configuration properties and a simple API to send and receive deltas.
.NET - for C#
Your application class must implement:
•
•
GemStone::GemFire::Cache::IGFDelta
GemStone::GemFire::Cache::IGFSerializable
IGFDelta provides the methods, HasDelta, ToDelta, and FromDelta, which you program to report on,
send, and receive deltas for your class.
Additionally, for cloning, your class must implement the standard .NET IClonable interface and its Clone
method. See Cloning on page 242.
C++
Your application must publicly derive from:
•
•
gemfire::Delta
gemfire::Cacheable
Delta provides the methods, hasDelta, toDelta, fromDelta, which you program to report on, send,
and receive deltas for your class.
For cloning, use the clone method provided in the Delta interface. See Cloning on page 242.
Cloning
With cloning enabled, Pivotal GemFire does a deep copy of the object, using serialization. You can
improve performance by implementing the appropriate clone method for your API, making a deep copy of
anything to which a delta may be applied.
With cloning enabled, Pivotal GemFire does a deep copy of the object, using serialization. You can
improve performance by implementing the appropriate clone method for your API, making a deep copy of
anything to which a delta may be applied.
The goal is to significantly reduce the overhead of copying the object while still retaining the isolation
needed for your deltas.
You configure delta propagation on the server side as well as client. For information on the server and
delta propagation, see the Delta Propagation chapter in the GemFire User’s Guide.
cloning-enabled
The cloning-enabled property is a region attributes boolean, configured in the cache.xml, that affects
how fromDelta applies deltas to the local client cache. When true, the updates are applied to a clone
of the value and then the clone is saved to the cache. When false, the value is modified in place in the
cache. The default value is false.
Cloning can be expensive, but it ensures that the new object is fully initialized with the delta before any
application code sees it.
Without cloning:
242
Implementing Delta Propagation
•
•
•
Pivotal GemFire Native Client
It is possible for application code to read the entry value as it is being modified, possibly seeing the
value in an intermediate, inconsistent state, with just part of the delta applied. You may choose to
resolve this issue by having your application code synchronize on reads and writes.
GemFire loses any reference to the old value because the old value is transformed in place into
the new value. Because of this, your CacheListener sees the same new value returned for
EntryEvent.getOldValue and EntryEvent.getNewValue.
Exceptions thrown from fromDelta may leave your cache in an inconsistent state. Without cloning,
any interruption of the delta application could leave you with some fields in your cached object changed
and others unchanged. If you do not use cloning, keep this in mind when you program your error
handling in your fromDelta implementation.
Enabling Cloning in cache.xml
<region name="exampleRegion">
<region-attributes refid="CACHING_PROXY" cloningenabled="true"
pool-name="examplePool"/>
</region>
Enabling Cloning (C++)
RegionFactoryPtr regionFactory =
cachePtr->createRegionFactory(CACHING_PROXY);
RegionPtr regionPtr = regionFactory
->setCloningEnabled(true)
->create("myRegion");
Implementing Delta Propagation
By default, delta propagation is enabled in your distributed system and is used for objects that implement
the delta interface. You program the client-side methods to extract delta information for your entries and to
apply received delta information.
By default, delta propagation is enabled in your distributed system and is used for objects that implement
the delta interface. You program the client-side methods to extract delta information for your entries and to
apply received delta information.
For more information, see Delta Propagation API.
Prerequisites
•
•
•
Study your object types and expected application behavior to determine which objects should use
delta propagation. Delta propagation is not beneficial for all data and data modification scenarios. See
Exceptions and Limitations on page 244.
Decide whether to enable cloning. Cloning is disabled by default. See cloning-enabled.
• If you enable cloning, consider providing your own implementation, to optimize performance.
• If you do not enable cloning, be sure to synchronize your delta code.
If you do not enable cloning, review all associated listener code for dependencies on the entry event
old value. Without cloning, GemFire modifies the entry in place and so loses its reference to the old
value. For delta events, the EntryEvent methods to retrieve the old and new values both return the
new value.
Procedure
243
Exceptions and Limitations
Pivotal GemFire Native Client
For every class in which you want delta propagation, implement the delta interface and update your
methods to support delta propagation. Exactly how you do this depends on your application and object
needs, but these steps describe the basic approach.
1. Study the object contents to decide how to handle delta changes.
Delta propagation has the same issues of distributed concurrency control as the distribution of full
objects, but on a more detailed level. Some parts of your objects may be able to change independent
of one another while others may always need to change together. Send deltas large enough to keep
your data logically consistent. If, for example, field A and field B depend on each other, then your delta
distributions should either update both fields or neither. As with regular updates, the fewer producers
you have on a data region, the lower your likelihood of concurrency issues.
2. In the application code that puts entries, put the fully populated object into the local cache. This
usually means doing a get on the entry, unless you are sure it does not already exist anywhere in the
distributed region.
Even though you are planning to send only deltas, errors on the receiving end could cause GemFire
to request the full object, so you must provide it to the originating put method. Do this even in empty
producers, with regions configured for no local data storage.
3. Change each field’s update method to record information about the update.
4.
5.
6.
7.
8.
The information must be sufficient for toDelta to encode the delta and any additional required delta
information when it is invoked.
Write hasDelta to report on whether a delta is available.
When writing your serialization methods, toDelta, fromDelta, toData, fromData, be sure to
exclude any fields used to manage delta state, which do not need to be sent.
Write toDelta to create a byte stream with the changes to the object and any other information that
fromDelta will need to apply the changes. Before returning from toDelta, reset your delta state to
indicate that there are no delta changes waiting to be sent.
Write fromDelta to decode the byte stream that toDelta creates and update the object.
Make sure you provide adequate synchronization to your object to maintain a consistent object state.
If you do not use cloning, you will probably need to synchronize on reads and writes to avoid reading
partially written updates from the cache. This might involve toDelta, fromDelta, and other object
access and update methods. Additionally, your implementation should take into account the possibility
of concurrent invocations of fromDelta and one or more of the object's update methods.
Exceptions and Limitations
In some application scenarios, delta propagation does not show any significant performance benefits. On
occasion it results in degradation. There are other limitations and exceptions related to delta propagation.
In some application scenarios, delta propagation does not show any significant performance benefits. On
occasion it results in degradation. There are other limitations and exceptions related to delta propagation.
•
•
•
•
Errors in Delta Propagation on page 244
Performance Limitations on page 245
Configurations That Require Full Values on page 245
General Limitations on page 246
Errors in Delta Propagation
Errors in delta propagation fall into two categories based on how they are handled:
•
Problems applying the delta that can be remedied if the originating member sends the full value
in place of the delta. Your put operation does not see errors or exceptions related to the failed delta
244
Delta Propagation
Pivotal GemFire Native Client
propagation. The system automatically does a full value distribution to the receiver where the problem
occurs. This type of error includes:
•
•
Unavailable entry value in the receiving cache, either because the entry is missing or its value is
null. In both cases, there is nothing to apply the delta to and the full value must be sent. This is
most likely to occur if you destroy or invalidate your entries locally, either through application calls or
through configured actions like eviction or entry expiration.
• InvalidDeltaException thrown by fromDelta method, programmed by you. This exception
enables you to avoid applying deltas that would violate data consistency checks or other application
requirements. Throwing this exception causes GemFire to send the full value.
• Any error applying the delta in a client in server-to-client propagation. The client retrieves the full
value from the server.
Problems creating or distributing the delta that cannot be fixed by distributing the full value.
These problems are caused by errors or exceptions in hasDelta or toDelta. In these cases, your
put operation fails with an exception.
Performance Limitations
Consider the following situations in which delta propagation adversely affects performance:
•
•
•
•
Added costs of deserializing your objects to apply deltas. Applying a delta requires the entry value
to be deserialized. Once this is done, the object is stored back in the cache in deserialized form. This
aspect of delta propagation only negatively impacts your system if your objects are not already being
deserialized for other reasons, such as for indexing and querying or for listener operations. Once stored
in deserialized form, there are reserialization costs for operations that send the object outside of the
member, like values sent in response to netSearch or client requests, and storage to disk. The more
operations that require reserialization, the higher the overhead of deserializing the object.
Cloning. Cloning can affect performance. Not using cloning is risky, however, as you are modifying
cached values in place. Make sure you synchronize your entry access to keep your cache from
becoming inconsistent.
Problems applying the delta that cause the system to go back to the originator for the full entry
value. In this scenario, the overall operation costs more than sending the full entry value in the first
place. The problem can be exacerbated if your delta is sent to a number of recipients, all or most of
them request a full value, and the full value send requires the object to be serialized.
Disk I/O costs associated with overflow regions. If you use eviction with overflow to disk, on-disk
values must be brought into memory in order to apply the delta. This is much more costly than removing
the reference to the disk copy, as you would do with a full value distribution into the cache.
Configurations That Require Full Values
GemFire clients can always send deltas to the servers, and servers can usually sent deltas to clients. The
following configurations require the servers to send full values to the clients, instead of deltas:
•
•
•
When the client's gemfire.properties setting conflate-events is set to true, the servers send
full values for all regions.
When the server region attribute enable-subscription-conflation is set to true and the client
gemfire.properties setting conflate-events is set to server, the servers send full values for
the region.
Servers send full values to client regions that are configured to not cache data—with the PROXY
RegionShortcut in their region attributes refid.
To use the delta propagation feature, all updates on a key in a region must have value types that
implement the Delta interface. You cannot mix object types for an entry key where some of the types
implement delta and some do not. This is because, when a type implementing the delta interface is
received for an update, the existing value for the key is cast to a Delta type to apply the received delta.
245
Examples of Delta Propagation
Pivotal GemFire Native Client
General Limitations
Sometimes fromDelta cannot be invoked because there is no object to apply the delta to in the receiving
cache. When this happens, the system sends the full value. There are two possible scenarios:
•
•
If the system can determine beforehand that the receiver does not have a local copy, it sends the initial
message with the full value. This is possible when regions are configured with no local data storage, as
when you are using them to provide data update information to listeners.
In less obvious cases, as when an entry has been locally deleted, first the delta is sent, then the
receiver requests a full value and that is sent. Whenever the full value is received, any further
distributions to the receiver’s peers or clients uses the full value.
GemFire does not propagate deltas for:
•
•
•
Transactional commit on the server
The putAll operation
Server VMs running GemFire versions 6.0 and earlier
Examples of Delta Propagation
Examples describe delta propagation operations and provide implementation code for C# .NET and C++.
Examples describe delta propagation operations and provide implementation code for C# .NET and C++.
Delta Propagation in the Client/Server
In this example, the feeder client is connected to the first server, and the receiver client is connected to the
second. The servers are peers to each other.
Figure 14: Example of Delta Propagation in the Client/Server
These are the main operations shown in the example:
1. In the Feeder client, the application updates the entry object and puts the entry. In response to the put,
Pivotal GemFire calls hasDelta, which returns true.
2. GemFire calls toDelta and forwards the extracted delta to the server. If hasDelta returned false,
GemFire would distribute the full entry value.
246
Delta Propagation
Pivotal GemFire Native Client
3. In Server1, GemFire applies the delta to the cache, distributes the received delta to the server’s peers,
and forwards it to any other clients with interest in the entry (there are no other clients to Server1 in this
example).
4. In Server2, GemFire applies the delta to the cache and forwards it to its interested clients, which in this
case is just Receiver client.
Client Example Files
These example files, from the product quickstart examples, show the basic approach to programming a
delta propagation implementation for an entry value object, named DeltaExample in this example.
XML file used for the examples
<cache>
<region name="root" refid="CACHING_PROXY">
<region-attributes cloning-enabled="true" pool-name="examplePool"/>
</region>
<pool name="examplePool" subscription-enabled="true" servergroup="ServerGroup1">
<locator host="localhost" port="34756"/>
</pool>
</cache>
Delta Example Implementation (C# .NET)
using System;
using GemStone.GemFire.Cache;
namespace GemStone.GemFire.Cache.QuickStart
{
public class DeltaExample : IGFDelta, IGFSerializable, ICloneable
{
// data members
private Int32 m_field1;
private Int32 m_field2;
private Int32 m_field3;
// delta indicators
private bool m_f1set;
private bool m_f2set;
private bool m_f3set;
public DeltaExample(Int32 field1, Int32 field2, Int32 field3)
{
m_field1 = field1;
m_field2 = field2;
m_field3 = field3;
reset();
}
public DeltaExample()
{
reset();
}
public DeltaExample(DeltaExample copy)
{
m_field1 = copy.m_field1;
m_field2 = copy.m_field2;
m_field3 = copy.m_field3;
reset();
}
247
Delta Propagation
Pivotal GemFire Native Client
private void reset()
{
m_f1set = false;
m_f2set = false;
m_f3set = false;
}
public Int32 getField1()
{
return m_field1;
}
// REPEAT FOR OTHER FIELDS
public void setField1(Int32 val)
{
lock(this)
{
m_field1 = val;
m_f1set = true;
}
}
// REPEAT FOR OTHER FIELDS
public bool HasDelta()
{
return m_f1set || m_f2set || m_f3set;
}
public void ToDelta(DataOutput DataOut)
{
lock(this)
{
DataOut.WriteBoolean(m_f1set);
if (m_f1set)
{
DataOut.WriteInt32(m_field1);
}
// REPEAT FOR OTHER FIELDS
}
}
reset();
public void FromDelta(DataInput DataIn)
{
lock(this)
{
m_f1set = DataIn.ReadBoolean();
if (m_f1set)
{
m_field1 = DataIn.ReadInt32();
}
// REPEAT FOR OTHER FIELDS
}
}
public void ToData(DataOutput DataOut)
{
DataOut.WriteInt32(m_field1);
DataOut.WriteInt32(m_field2);
DataOut.WriteInt32(m_field3);
}
248
Delta Propagation
Pivotal GemFire Native Client
public IGFSerializable FromData(DataInput DataIn)
{
m_field1 = DataIn.ReadInt32();
m_field2 = DataIn.ReadInt32();
m_field3 = DataIn.ReadInt32();
return this;
}
public UInt32 ClassId
{
get
{
return 0x02;
}
}
public UInt32 ObjectSize
{
get
{
UInt32 objectSize = 0;
return objectSize;
}
}
public static IGFSerializable create()
{
return new DeltaExample();
}
}
}
public Object Clone()
{
return new DeltaExample(this);
}
Delta Example Implementation (C++)
#ifndef __Delta_Example__
#define __Delta_Example__
#include <gfcpp/GemfireCppCache.hpp>
using namespace gemfire;
class DeltaExample: public Cacheable, public Delta
{
private:
// data
int32_t
int32_t
int32_t
members
m_field1;
m_field2;
m_field3;
// delta indicators
mutable bool m_f1set;
mutable bool m_f2set;
mutable bool m_f3set;
public:
249
Delta Propagation
Pivotal GemFire Native Client
DeltaExample(int32_t field1, int32_t field2, int32_t field3) :
m_field1(field1), m_field2(field2), m_field3(field3)
{
reset();
}
DeltaExample()
{
reset();
}
DeltaExample(DeltaExample * copy)
{
m_field1 = copy->m_field1;
m_field2 = copy->m_field2;
m_field3 = copy->m_field3;
reset();
}
void reset() const
{
m_f1set = false;
m_f2set = false;
m_f3set = false;
}
int getField1()
{
return m_field1;
}
// REPEAT FOR OTHER FIELDS
void setField1(int val)
{
lock();
m_field1 = val;
m_f1set = true;
unlock();
}
// REPEAT FOR OTHER FIELDS
virtual bool hasDelta()
{
return m_f1set || m_f2set || m_f3set;
}
virtual void toDelta(DataOutput& out) const
{
lock();
out.writeBoolean(m_f1set);
if (m_f1set)
{
out.writeInt(m_field1);
}
// REPEAT FOR OTHER FIELDS
reset();
}
unlock();
virtual void fromDelta(DataInput& in)
{
250
Delta Propagation
Pivotal GemFire Native Client
lock();
in.readBoolean(&m_f1set);
if (m_f1set)
{
in.readInt(&m_field1);
}
// REPEAT FOR OTHER FIELDS
reset();
}
unlock();
virtual void toData(DataOutput& output) const
{
lock();
output.writeInt(m_field1);
output.writeInt(m_field2);
output.writeInt(m_field3);
unlock();
}
virtual Serializable* fromData(DataInput& input)
{
lock();
input.readInt(&m_field1);
input.readInt(&m_field2);
input.readInt(&m_field3);
unlock();
return this;
}
virtual int32_t classId() const
{
return 2;
}
virtual uint32_t objectSize() const
{
return 0;
}
DeltaPtr clone()
{
return DeltaPtr(new DeltaExample(this));
}
virtual ~DeltaExample()
{
}
static Serializable* create()
{
return new DeltaExample();
}
void lock() const { /* add your platform dependent syncronization code
here */ }
void unlock() const { /* add your platform dependent syncronization code
here */ }
};
251
Delta Propagation
Pivotal GemFire Native Client
#endif
252
Delta Propagation
Pivotal GemFire Native Client
253
Appendix A: Interoperability of Language Classes and Types
Pivotal GemFire Native Client
Part
XVI
Appendix A: Interoperability of Language Classes
and Types
Topics:
•
•
C++ Class to .NET Class
Mappings
Java to .NET Type Mapping
Table
Interoperability of Language Classes and Types provides a table that
maps C++ class methods to corresponding .NET class methods and a
table that maps Java types to .NET types.
Interoperability of Language Classes and Types provides a table that
maps C++ class methods to corresponding .NET class methods and a
table that maps Java types to .NET types.
254
C++ Class to .NET Class Mappings
Pivotal GemFire Native Client
C++ Class to .NET Class Mappings
Wherever the native C++ class methods use pass-by-reference semantics to return data, the
corresponding .NET methods shown in the following table return the object instead of using pass-byreference semantics.
Table 24: C++ Class to .NET Class Mappings
C++ Class
.NET Class
class gemfire::AttributesFactory
Sealed class AttributesFactory
class gemfire::AttributesMutator
Sealed class AttributesMutator
class gemfire::Cache
Sealed class Cache
abstract class gemfire:: Cacheable
Interface IPdxSerializable or interface
IGFSerializable
class gemfire::CacheableBytes
Byte[] or ArrayList<Byte>
class gemfire::Cacheableint32
Int32
class gemfire:: CacheableString
String
abstract class gemfire:: CacheableKey
You can use any type that implements hashcode
and equals. The generic .NET built-in types are all
suitable.
abstract class gemfire::CacheListener
Interface ICacheListener
class gemfire:: CacheLoader
Interface ICacheLoader plus static class
CacheLoader
class gemfire:: CacheWriter
Interface class ICacheWriter
class gemfire::CacheFactory
Sealed class CacheFactory
class gemfire::DataInput
With IPdxSerializable, IPdxReader.
With IGFSerializable, sealed class
DataInput.
class gemfire:: DataOutput
With IPdxSerializable, IPdxWriter.
With IGFSerializable, sealed class
DataOutput.
class gemfire:: DiskPolicyType
enum DiskPolicyType plus static class
DiskPolicy containing convenience methods for
DiskPolicyType enumeration
class gemfire:: DistributedSystem
Sealed class DistributedSystem
255
Appendix A: Interoperability of Language Classes and Types
Pivotal GemFire Native Client
C++ Class
.NET Class
class gemfire:: EntryEvent
Sealed class EntryEvent
class gemfire:: Exception
Class GemfireException
all other exceptions deriving from gemfire::
Exception
Corresponding exceptions deriving from
GemfireException
class gemfire:: ExpirationAction
enum ExpirationAction plus static class
Expiration containing convenience methods for
ExpirationAction enumeration
class gemfire:: Log
Static class Log. The native Log::log method is
mapped to Log.Write to avoid the conflict with the
class name which is reserved for the constructors
of Log class. The various loglevel Throw or Catch
methods are not implemented, since they are
redundant to Log::Log , Log::LogThrow, and
Log::LogCatch methods that take LogLevel as
a parameter.
enum gemfire:: MemberType
enum MemberType
abstract class gemfire::
PersistanceManager
Not provided. You can register
a C++ implementation using
AttributesFactory.SetPersistenceManager
but you cannot implement a new one in .NET
class gemfire:: Properties
Sealed class Properties
class gemfire:: Properties::Visitor
Delegate PropertiesVisitor
abstract class gemfire:: Region
Class IRegion
class gemfire:: RegionAttributes
Sealed class RegionAttributes
class gemfire:: ScopeType
enum ScopeType plus static class Scope
containing convenience methods for ScopeType
enumeration+
abstract class gemfire:: Serializable
Two options:
•
•
Interface IPdxSerializable
Interface IGFSerializable plus
wrapper Serializable class for native
Serializable and UserData objects. The
native toString method is not provided, since
the ToString method of the base object class
provides the same functionality.
class gemfire:: SystemProperties
Sealed class SystemProperties
class gemfire:: UserData
Two options:
256
Appendix A: Interoperability of Language Classes and Types
C++ Class
Pivotal GemFire Native Client
.NET Class
•
•
class gemfire:: VectorT<T>
Interface IPdxSerializable
Interface IGFSerializable
Array of the given type, such as T[]
Interoperability of C++ Types When Using PDX
Serialization
This topic table lists the mapping between C++ types and other language types when using PDX
serialization.
In addition, the table lists which PdxReader and PdxWriter C++ APIs to use when serializing and
deserializing the types.
C++ Type
.NET Type
CacheableHashTable
System::Collections::Hashtable
java.util.Hashtable
CacheableHashMap
System::Collections ::Generic::IDictionary<Object,
java.util.HashMap
readObject/writeObject
Object>
CacheableVector
System::Collections::ArrayList
java.util.Vector
readObject/writeObject
CacheableArrayList
System::Collections:: Generic::IList<Object>
java.util.ArrayList
readObject/writeObject
bool
bool
boolean
readBoolean/
writeBoolean
int8_t
sbyte
Byte
readByte/writeByte
Char
Char
readChar/writeChar
wchar_t*/char*
string
string
readString/writeString
double
Double
double
readDouble/writeDouble
float
float
float
readFloat/writeFloat
int16_t
short
short
readShort/writeShort
int32_t
Int32/int
int
readInt/writeInt
int64_t
Int64/long
long
readLong/writeLong
int8_t*
System.Byte[]
(System.SByte[]??)
Byte[]
readByteArray/
writeByteArray
double*
System.Double[]
Double[]
readDoubleArray/
writeDoubleArray
float*
System.float[]
Float[]
readFloatArray/
writeFloatArray
CacheableHashSet
CacheableHashSet
java.util.HashSet
readObject/writeObject
CacheableLinkedHashSet
CacheableLinkedHashSetjava.util.LinkedHashSet
readObject/writeObject
wchar_t/char
1
1
Java Type
2
PdxReader/PdxWriter
API
readObject/writeObject
257
Java to .NET Type Mapping Table
Pivotal GemFire Native Client
C++ Type
.NET Type
Java Type
PdxReader/PdxWriter
API
int16_t*
System.Int16[]
Short[]
readShortArray/
writeShortArray
int32_t*
System.Int32[]
Int[]
readIntArray/
writeIntArray
int64_t*
System.Int64[]
Long[]
readLongArray/
writeLongArray
bool*
System.Boolean[]
Boolean[]
readBooleanArray/
writeBooleanArray
System.Char[]
char[]
readCharArray/
writeCharArray
enum
Enum
readObject/writeObject
byte[][]/Sbyte[][]
Byte[][]
readArrayOfByteArrays/
writeArrayOfByteArrays
System.String[]
String[]
readStringArray/
writeStringArray
Java.util.date
readDate/writeDate
1
wchar_t*/char*
enum
3
int8_t**
wchar_t**/char**
1
4
CacheableDate
System.DateTime
CacheableObjectArray
object[]/System.Object[] Object[]
readObjectArray/
writeObjectArray
Cacheable/Serializable
object/System.Object
readObject/writeObject
Object
1
C++ allows unicode and non-unicode characters, so C++ PDX will support both wchar_t/char and
wchar_t*/char*.
2
Using Pdx we use sByte only as java byte is signed. But for DataSerializable we have kept Byte[] array
only as this is use as data container.
3
C++ allows explicit setting of ordinal number, but it is up to the user to map java enumName with that of C
++ enumName. See Using C++ Enum Type with PDX Serialization on page 100.
4
Now we put UTC time from .NET .
Java to .NET Type Mapping Table
The following table provides a mapping between Java and .NET types.
Table 25: Java types and.NET types
Java Type
.NET Type
instances of PdxSerializable
.NET class of same name
instances of PdxInstance
.NET class of same name
instances serialized by a PdxSerializer
.NET class of same name
java.lang.Byte
System.SByte
258
Appendix A: Interoperability of Language Classes and Types
Pivotal GemFire Native Client
Java Type
.NET Type
java.lang.Boolean
System.Boolean
java.lang.Character
System.Char
java.lang.Short
System.Int16
java.lang.Integer
System.Int32
java.lang.Long
System.Int64
java.lang.Float
System.Float
java.lang.Double
System.Double
java.lang.String
System.String
java.util.Date
System.DateTime
byte[]
System.Byte[]
boolean[]
System.Boolean[]
char[]
System.Char[]
short[]
System.Int16[]
int[]
System.Int32[]
long[]
System.Int64[]
float[]
System.Float[]
double[]
System.Double[]
String[]
System.String[]
byte[][]
System.Byte[][]
Object[]
system.Collections.Generic.List<Object>
java.util.HashMap
System.Collections.Generics.IDictionary<Object,
Object>
java.util.Hashtable
System.Collections.Hashtable
java.util.ArrayList
System.Collections.Generic.IList<Object>
java.util.Vector
Collections.ArrayList
java.util.HashSet
CacheableHashSet
259
Appendix A: Interoperability of Language Classes and Types
Pivotal GemFire Native Client
Java Type
.NET Type
java.util.LinkedHashSet
CacheableLinkedHashSet
260
Appendix A: Interoperability of Language Classes and Types
Pivotal GemFire Native Client
261
Appendix B: System Statistics
Pivotal GemFire Native Client
Part
XVII
Appendix B: System Statistics
Topics:
•
•
•
Sampling Statistics
System Performance Statistics
Operating System Statistics
System Statistics provides information on the Pivotal GemFire
installation and includes standard statistics for caching and distribution
activities.
System Statistics provides information on the Pivotal GemFire
installation standard statistics for caching and distribution activities.
Statistics that end with "time" are time-based statistics. For
performance reasons, by default, the system does not collect
these. To enable time-based statistics gathering, set the system
property enable-time-statistics in Table 9: Attributes in
gfcpp.properties: Statistics Archiving Properties on page 41.
262
Sampling Statistics
Pivotal GemFire Native Client
Sampling Statistics
When applications and cache servers join a distributed system, they indicate whether to enable statistics
sampling and whether to archive the statistics that are gathered.
The following statistics are related to the statistic sampler.
sampleCount
Total number of samples taken by this sampler.
sampleTime
Total amount of time spent taking samples.
StatSampler
Statistics on the statistic sampler.
For more information about configuring statistics, see Configuration Options on page 39.
System Performance Statistics
Performance statistics are collected for each application or cache server that connects to a distributed
system.
Region Statistics
These methods help you to get the statistics of a region.
creates
Total number of cache creates for this region.
puts
Total number of cache put operations for this region.
putTime
Total time spent doing put operations for this region.
putAll
Total number of cache putAlls for this region.
putAllTime
Total time spent doing putAll operations for this region.
gets
Total number of cache gets for this region.
getTime
Total time spent doing get operations for this region.
getAll
Total number of cache getAlls for this region.
getAllTime
Total time spent doing getAll operations for this region.
hits
Total number of cache hits for this region.
misses
Total number of cache misses for this region.
entries
Current number of cache entries for this region.
destroys
Total number of cache destroys for this region.
clears
Total number of cache clears for this region.
overflows
Total number of cache overflows to disk for this region.
retrieves
Total number of cache entries fetched from disk into the cache region.
nonSingleHopCount
Total number of times client request required multiple hops.
263
Appendix B: System Statistics
Pivotal GemFire Native Client
metaDataRefreshCount
Total number of times metadata was refreshed due to the observation
of multiple hops.
cacheLoaderCallCompleted
Total number of times a load has completed for this region.
cacheLoaderCallTime
Total time spent invoking the loaders for this region.
CacheWriterCallsCompleted Total number of times a cache writer call has completed for this
region.
CacheWriterCallTime
Total time spent doing cache writer calls.
CacheListenerCallsCompleted
Total number of times a cache listener call has completed for this
region.
CacheListenerCallTime
Total time spent doing cache listener calls for this region.
Cache Performance Statistics
Use cache performance statistics to determine the type and number of cache operations being performed
and how much time they consume.
These statistics are available if the member creates a cache.
creates
Total number of cache creates.
puts
Total number of cache puts.
gets
Total number of cache gets.
entries
Current number of cache entries.
hits
Total number of cache hits.
misses
Total number of cache misses.
destroys
Total number of cache destroys.
overflows
Total number of cache overflows to persistence
backup.
cacheListenerCallsCompleted
Total number of times a cache listener call has
completed.
pdxInstanceDeserializations
Total number of times getObject has been called on
a PdxInstance.
pdxInstanceDeserializationTime
Total amount of time, in nanoseconds, spent
deserializing PdxInstances by calling getObject.
pdxInstanceCreations
Total number of times a deserialization created a
PdxInstance.
pdxSerializations
Total number of PDX serializations.
pdxSerializedBytes
Total number of bytes produced by PDX
serialization.
pdxDeserializations
Total number of PDX deserializations.
pdxDeserializedBytes
Total number of bytes read by PDX deserialization.
tombstoneCount
Total number of tombstone entries created for
performing concurrency checks.
264
Appendix B: System Statistics
nonReplicatedTombstoneSize
Pivotal GemFire Native Client
Approximate total size (in bytes) of tombstones
present in the client cache.
Continuous Query Statistics
Continuous query statistics give you information about a registered Continuous Query (CQ) represented by
the CqQuery object.
inserts
Total number of inserts for this CQ query.
updates
Total number of updates for this CQ query.
deletes
Total number of deletes for this CQ query.
events
Total number of events for this CQ query.
CQ Service Statistics
Use CQ service methods to get aggregate statistical information about the continuous queries of a client.
CqsActive
Total number of CQs active for this CQ service.
CqsC CqsActive reated
Total number of CQs created for this CQ service.
CqsClosed
Total number of CQs closed for this CQ service.
CqsStopped
Total number of CQs stopped for this CQ service.
CqsOnClient
Total number of CQs on the client for this CQ
service.
Pool Statistics
Use the pool object to get statistics on connection pools.
locators
Current number of locators discovered.
servers
Current number of servers discovered.
? servers ?
Number of servers hosting this clients
subscriptions.
locatorRequests
Number of requests from this connection pool to a
locator.
locatorResponses
Number of responses from the locator to this
connection pool.
poolConnections
Current number of pool connections.
connects
Total number of times a connection has been
created.
ConnectionWaitTime
Total time (nanoseconds) spent waiting for a
connection.
disconnects
Total number of times a connection has been
destroyed.
265
Appendix B: System Statistics
Pivotal GemFire Native Client
minPoolSizeConnect
Total number of connects done to maintain
minimum pool size.
loadConditioningConnects
Total number of connects done due to load
conditioning.
idleDisconnects
Total number of disconnects done due to idle
expiration.
loadConditioningDisconnects
Total number of disconnects done due to load
conditioning expiration.
connectionWaitsInProgress
Current number of threads waiting for a connection.
connectionWaits
Total number of times a thread completed waiting
for a connection (by timing out or by getting a
connection).
clientOpsInProgress
Current number of clientOps being executed.
clientOps
Total number of clientOps completed successfully.
clientOpFailures
Total number of clientOp attempts that have failed.
clientOpTimeouts
Total number of clientOp attempts that have timed
out.
QueryExecutions
Total number of queryExecutions.
QueryExecutionTime
Total time spent while processing queryExecution.
Delta Statistics
Delta statistics provide information about updates to data.
deltaMessageFailures
Total number of messages containing delta
(received from server) but could not be processed
after reception.
deltaPuts
Total number of puts containing delta that have
been sent from client to server.
processedDeltaMessages
Total number of messages containing delta
received from server and processed after reception.
processedDeltaMessagesTime
Total time spent applying delta (received from
server) on existing values at client.
266
Operating System Statistics
Pivotal GemFire Native Client
Operating System Statistics
Use operating system statistics to determine a member’s CPU, memory, and disk usage.
Linux Process Statistics
Use these methods to get information about a Linux operating system process that is using a GemFire
system.
imageSize
The size of the process's image in megabytes.
rssSize
The size of the process's resident set size in
megabytes.
userTime
The operating system statistic for the process CPU
usage in user time
systemTime
The operating system statistic for the process CPU
usage in system time.
hostCpuUsage
The operating system statistic for the host CPU
usage.
threads
Number of threads currently active in this process.
LinuxProcessStats
Statistics for a Linux process.
Solaris Process Statistics
Use these methods to get information about a Solaris operating system process that is using a GemFire
system
imageSize
Size of the process image in megabytes.
rssSize
Size of the process resident set in megabytes.
userTime
Operating system statistic for the process CPU
usage in user time.
systemTime
Operating system statistic for the process CPU
usage in system time.
processCpuUsage
Operating system statistic for the CPU usage of this
process.
hostCpuUsage
Operating system statistic for the host CPU usage.
threads
Number of threads currently active in this process.
267
Appendix B: System Statistics
Pivotal GemFire Native Client
Windows Process Statistics
Use these methods to get information about a Windows operating system process that is using a GemFire
system.
handles
Total number of handles currently open by this
process. This number is the sum of the handles
currently open by each thread in this process.
priorityBase
Current base priority of the process. Threads
within a process can raise and lower their own
base priority relative to the process's base priority.
threads
Number of threads currently active in this process.
An instruction is the basic unit of execution in
a processor, and a thread is the object that
executes instructions. Every running process has
at least one thread.
activeTime
Elapsed time in milliseconds that all threads
of this process used the processor to execute
instructions. An instruction is the basic unit of
execution in a computer, a thread is the object
that executes instructions, and a process is the
object created when a program is run. Code
executed to handle some hardware interrupts and
trap conditions are included in this count.
pageFaults
Total number of Page Faults by the threads
executing in this process. A page fault occurs
when a thread refers to a virtual memory page
that is not in its working set in main memory. This
will not cause the page to be fetched from disk if
it is on the standby list and hence already in main
memory, or if it is in use by another process with
whom the page is shared.
pageFileSize
Current number of bytes this process has used in
the paging file(s). Paging files are used to store
pages of memory used by the process that are not
contained in other files. Paging files are shared
by all processes, and lack of space in paging
files can prevent other processes from allocating
memory.
pageFileSizePeak
Maximum number of bytes this process has used
in the paging file(s). Paging files are used to store
pages of memory used by the process that are not
contained in other files. Paging files are shared
by all processes, and lack of space in paging
files can prevent other processes from allocating
memory.
privateSize
Current number of bytes this process has
allocated that cannot be shared with other
processes.
268
Appendix B: System Statistics
Pivotal GemFire Native Client
systemTime
Elapsed time in milliseconds that the threads
of the process have spent executing code in
privileged mode. When a Windows system service
is called, the service often runs in Privileged Mode
to gain access to system-private data. Such data
is protected from access by threads executing in
user mode. Calls to the system can be explicit or
implicit, such as page faults or interrupts. Unlike
some early operating systems, Windows uses
process boundaries for subsystem protection in
addition to the traditional protection of user and
privileged modes. These subsystem processes
provide additional protection. Therefore, some
work done by Windows on behalf of your
application might appear in other subsystem
processes in addition to the privileged time in your
process.
userTime
Elapsed time in milliseconds that this process's
threads have spent executing code in user
mode. Applications, environment subsystems,
and integral subsystems execute in user mode.
Code executing in User Mode cannot damage
the integrity of the Windows Executive, Kernel,
and device drivers. Unlike some early operating
systems, Windows uses process boundaries
for subsystem protection in addition to the
traditional protection of user and privileged
modes. These subsystem processes provide
additional protection. Therefore, some work done
by Windows on behalf of your application might
appear in other subsystem processes in addition
to the privileged time in your process.
activeTime
Elapsed time in milliseconds that all threads
of this process used the processor to execute
instructions. An instruction is the basic unit of
execution in a computer, a thread is the object
that executes instructions, and a process is the
object created when a program is run. Code
executed to handle some hardware interrupts and
trap conditions are included in this count.
pageFaults
Total number of page faults by the threads
executing in this process. A page fault occurs
when a thread refers to a virtual memory page
that is not in its working set in main memory. This
will not cause the page to be fetched from disk if
it is on the standby list and hence already in main
memory, or if it is in use by another process with
whom the page is shared.
pageFileSize
Current number of bytes this process has used in
the paging file(s). Paging files are used to store
pages of memory used by the process that are not
contained in other files. Paging files are shared
by all processes, and lack of space in paging
269
Appendix B: System Statistics
Pivotal GemFire Native Client
files can prevent other processes from allocating
memory.
pageFileSizePeak
Maximum number of bytes this process has used
in the paging file(s). Paging files are used to store
pages of memory used by the process that are not
contained in other files. Paging files are shared
by all processes, and lack of space in paging
files can prevent other processes from allocating
memory.
privateSize
Current number of bytes this process has
allocated that cannot be shared with other
processes.
systemTime
Elapsed time in milliseconds that the threads
of the process have spent executing code in
privileged mode. When a Windows system service
is called, the service often runs in privileged mode
to gain access to system-private data. Such data
is protected from access by threads executing in
user mode. Calls to the system can be explicit or
implicit, such as page faults or interrupts. Unlike
some early operating systems, Windows uses
process boundaries for subsystem protection in
addition to the traditional protection of user and
privileged modes. These subsystem processes
provide additional protection. Therefore, some
work done by Windows on behalf of your
application might appear in other subsystem
processes in addition to the privileged time in your
process.
userTime
Elapsed time in milliseconds that this process's
threads have spent executing code in user
mode. Applications, environment subsystems,
and integral subsystems execute in user mode.
Code executing in User Mode cannot damage
the integrity of the Windows Executive, Kernel,
and device drivers. Unlike some early operating
systems, Windows uses process boundaries
for subsystem protection in addition to the
traditional protection of user and privileged
modes. These subsystem processes provide
additional protection. Therefore, some work done
by Windows on behalf of your application might
appear in other subsystem processes in addition
to the privileged time in your process.
virtualSize
Current size in bytes of the virtual address space
the process is using. Use of virtual address space
does not necessarily imply corresponding use of
either disk or main memory pages. Virtual space
is finite, and by using too much, the process can
limit its ability to load libraries.
virtualSizePeak
Maximum number of bytes of virtual address
space the process has used at any one time. Use
of virtual address space does not necessarily
270
Appendix B: System Statistics
Pivotal GemFire Native Client
imply corresponding use of either disk or main
memory pages. Virtual space is however finite,
and by using too much, the process might limit its
ability to load libraries.
workingSetSize
Current number of bytes in the Working Set of this
process. The Working Set is the set of memory
pages touched recently by the threads in the
process. If free memory in the computer is above
a threshold, pages are left in the Working Set
of a process even if they are not in use. When
free memory falls below a threshold, pages are
trimmed from Working Sets. If pages are needed,
they are then soft-faulted back into the Working
Set before they are paged out to disk.
workingSetSizePeak
Maximum number of bytes in the Working Set of
this process at any point in time. The Working Set
is the set of memory pages touched recently by
the threads in the process. If free memory in the
computer is above a threshold, pages are left in
the Working Set of a process even if they are not
in use. When free memory falls below a threshold,
pages are trimmed from Working Sets. If they are
needed they will then be soft faulted back into the
Working Set before they leave main memory.
cpuUsage
Percentage CPU used by this process.
WindowsProcessStats
Statistics for a Microsoft Windows process.
271
Appendix C: Installing the SQLite Persistence Manager
Pivotal GemFire Native Client
Part
XVIII
Appendix C: Installing the SQLite Persistence
Manager
Topics:
•
•
•
Linux Installation
Solaris Installation
Windows Installation
Installing the SQLite Persistence Manager describes how to download,
build and install the SQLite database libraries for use with disk
overflow.
Installing the SQLite Persistence Manager describes how to download,
build and install the SQLite database libraries for use with disk
overflow.
The commands that you type are shown in boldface fixed font. See
PersistenceManager on page 69 for additional information about the
SQLite database libraries.
272
Linux Installation
Pivotal GemFire Native Client
Linux Installation
This topic describes how to install the SQLite Persistence Manager on Linux for use with the GemFire
native client.
The productDir directory refers to the path to the native client product directory.
The following libraries must be present in the runtime linking path:
•
•
libSqLiteImpl.so is provided in productDir/lib, so it is already present in the runtime linking
path.
libsqlite3.so is the SQLite Library. You need to create this library and make available in the
runtime linking path, or copied to productDir/lib, as described below.
Pivotal GemFire Native Client has been tested with SQLite version 3.7.14.1.
Downloading, Building and Installing the Library
You create the SQLite database library by downloading the latest .zip file and compiling the source code.
1. Download the source code sqlite-autoconf-NNNNNNN.tar.gz file (where NNNNNNN
corresponds to the version) for SQLite version 3.7.14.1 or later from http://www.sqlite.org/
download.html.
2. Extract the source code from the .tar.gz file. For example:
tar -xvf sqlite-autoconf-3071401.tar.gz
3. Change directories to the extracted source files, and follow the install instructions located in the
"INSTALL" file.
a. Run the configure command for 32-bit or 64-bit with the following options, all entered on a single
command line. Change the --prefix directory specification to the location where you want the
libraries:
•
32-bit:
CFLAGS="-m32" ./configure --prefix=/desired_binary_location/sqlitebinaries
•
64-bit:
./configure --prefix=/desired_binary_location/sqlite-binaries
b. Run gmake install as described in the build instructions. The libraries will be available in the
sqlite-binaries directory that you specified.
4. Copy /desired_binary_location/sqlite-binaries/lib/libsqlite3.so file to
productDir/lib.
Solaris Installation
This topic describes how to install the SQLite Persistence Manager on Solaris for use with the GemFire
native client.
The productDir directory refers to the path to the native client product directory.
The following libraries must be present in the runtime linking path:
273
Windows Installation
•
•
Pivotal GemFire Native Client
libSqLiteImpl.so is provided in productDir/lib, so it is already present in the runtime linking
path.
libsqlite3.so is the SQLite Library. You need to create this library and make available in the
runtime linking path, or copied to productDir/lib, as described below.
Pivotal GemFire Native Client has been tested with SQLite version 3.7.14.1.
Downloading, Building, and Installing the Library
You create the SQLite database library by downloading the latest .zip file and compiling the source code.
1. Download the source code sqlite-autoconf-NNNNNNN.tar.gz file (where NNNNNNN
corresponds to the version) for SQLite version 3.7.14.1 or later from http://www.sqlite.org/
download.html.
2. Update your PATH environment variable to include the location of the Solaris ar command.
export PATH=/usr/css/bin:$PATH
3. Extract the source code from the .tar.gz file. First unzip:
gzip -d sqlite-autoconf-3071401.tar.gz
Then untar the file:
tar -xvf sqlite-autoconf-3071401.tar
4. Change directories to the extracted source files, and follow the install instructions located in the
"INSTALL" file.
a. Run the configure command for 32-bit or 64-bit Solaris systems with the following options, all
entered on a single command line. Change the --prefix directory specification to the location
where you want the libraries:
•
32-bit:
CC=cc CFLAGS="-xarch=v8plus -code=pic32" ./configure -prefix=/desired_binary_location/sqlite-binaries
•
64-bit:
CC=cc CFLAGS="-xarch=v9 -code=pic32" ./configure -prefix=/desired_binary_location/sqlite-binaries CFLAGS="-m64"
b. Run gmake install. The libraries will be available in the sqlite-binaries directory that you
specified.
5. Copy /desired_binary_location/sqlite-binaries/lib/libsqlite3.so file to
productDir/lib.
Windows Installation
This topic describes how to install the SQLite Persistence Manager on Windows for use with the Pivotal
GemFire native client.
Pivotal GemFire Native Client has been tested with SQLite version 3.7.14.1.
The productDir directory refers to the native client product directory path.
The following libraries are required. The productDir\bin directory containing these libraries must be
present in the Windows PATH environment variable, and that directory is added to PATH during the Pivotal
GemFire product installation.
274
Appendix C: Installing the SQLite Persistence Manager
•
•
•
Pivotal GemFire Native Client
The sqliteimpl.dll and GemStone.GemFire.Plugins.SQLite.dll files are provided in
productDir\bin.
For .NET C# native client application development, you need to obtain the
System.Data.SQLite.dll SQLite library, as described below. The library can be copied to
productDir\bin.
For C++ native client application development, you need the SqLite3.dll SQLite Library. You
create this library and make it available in the runtime linking path, or copied to productDir\bin, as
described below.
Downloading Pre-built System.Data.SQLite.dll Binaries
If you are writing native client applications using the .NET caching API, obtain the SQLite library (version
3.7.14.1 or later) for Windows as follows:
1. Access the System.Data.SQLite Download Page at the following URL: http://system.data.sqlite.org/
index.html/doc/trunk/www/downloads.wiki.
2. Download the appropriate setup file for your .NET Framework installation and hardware architecture.
For 32-bit Windows, under Setups for 32-bit Windows (.NET Framework 4.0) download sqlitenetFx40-setup-x86-2010-1.0.84.0.exe.
• For 64-bit Windows, under Setups for 64-bit Windows (.NET Framework 4.0) download sqlitenetFx40-setup-x64-2010-1.0.84.0.exe.
3. Execute the setup .exe file, and follow the prompts in the installation wizard. Accept all default
installation options.
4. Copy the C:\Program Files\System.Data.SQLite\2010\bin\System.Data.SQLite.dll
file to your Pivotal GemFire Native Client distribution at productDir\bin.
•
Downloading, Building, and Installing the Library
If you are writing native client applications using the C++ caching API, you need to build the SQLite
solution for your Windows platform architecture.
1. Download the source code sqlite-autoconf-NNNNNNN.tar.gz file (where NNNNNNN
corresponds to the version) for SQLite version 3.7.14.1 or later from http://www.sqlite.org/
download.html.
2. Extract the source code from the .tar.gz file. You may need to use CygWin or a Windows-compatible tar
extraction tool.
3. Using Visual Studio 2010, build the SqLite_3.7.14.1 solution either as a release or debug build:
• If you are using 32-bit Windows, build the SqLite_3.7.14.1, use the win32 configuration.
• If you are using 64-bit Windows, build the SqLite_3.7.14.1, use the x64 configuration.
4. From the built files, copy the SqLite3.dll file to your Pivotal GemFire Native Client distribution at
productDir\bin.
275
Glossary
Pivotal GemFire Native Client
Part
XIX
Glossary
This glossary defines terms used in Pivotal GemFire documentation.
API
Application Programming Interface. GemFire provides APIs to cached
data for C++ and .NET applications.
application program
A program designed to perform a specific function directly for the
user or, in some cases, for another application program. GemFire
applications use the GemFire application programming interfaces
(APIs) to modify cached data.
cache
A cache created by an application or cache server process. For the
process, its cache is the point of access to all caching features and
the only view of the cache that is available. Cache creation requires
membership in the distributed system. See also local cache and
remote cache.
cache configuration file
An XML file that declares the initial configuration of a cache, commonly
named cache.xml. C++ and .NET applications can configure the
cache additionally through the GemFire programming APIs.
cache listener
User-implemented plug-in for receiving and handling region entry
events. A region's cache listener is called after an entry in the local
cache is modified.
cache loader
User-implemented plug-in for loading data into a region. A region's
cache loader is used to load data that is requested of the region but
is not available in the distributed system. For a distributed region, the
loader that is used can be in a different cache from the one where
the data-request operation originated. See also cache writer and
netSearch.
276
Glossary
Pivotal GemFire Native Client
cache server
A long-running, configurable caching process, generally used to serve
cached data to the applications. Usually, cache servers are configured
to operate as servers in a client-server typology and their regions are
configured to be replicates. See also server.
cache writer
User-implemented plug-in intended for synchronizing the cache with
an outside data source. A region's cache writer is a synchronous
listener to cache data events. The cache writer has the ability to abort
a data modification. See also cache loader.
caching enabled
Specifies whether data is cached in the region. GemFire gives you the
option of running applications without entry caching. For example, you
can configure a distributed system as a simple messaging service.
client
In a client-server topology, clients can connect to cache servers,
create new regions on the cache server, and store data in the cache
server region. Clients can also connect to existing regions on a cache
server and do directed gets and puts on the cache server. Clients
do not track membership information about other clients, nor do they
share information with other clients.
concurrency level
An estimate of the number of threads expected to concurrently modify
values in the region. The actual concurrency may vary; this value is
used to optimize the allocation of system resources.
connection
What an application uses to access a GemFire distributed system.
An application can connect to a GemFire system by calling the
DistributedSystem::connect function with the appropriate
parameter settings. An application must connect to a distributed
system to gain access to GemFire functionality.
destroy
Remove an entry from a region or remove a region from a cache.
disk policy
Determines whether LRU entries exceeding the entries limit for a
caching region are destroyed or written to disk.
distributed scope
Enables a region to automatically send entry value updates to remote
caches and incorporate updates received from remote caches.
The scope identifies whether distribution operations must wait for
acknowledgement from other caches before continuing. A distributed
277
Glossary
Pivotal GemFire Native Client
region's cache loader and cache writer (defined in the local cache)
can be invoked for operations originating in remote caches.
distributed system
One or more GemFire system members that have been configured
to communicate with each other, forming a single, logical system.
Also used for the object that is instantiated to create the connection
between the distributed system members.
DTD
Document Type Definition. A language that describes the contents of a
Standard Generalized Markup Language (SGML) document. The DTD
is also used with XML. The DTD definitions can be embedded within
an XML document or in a separate file.
entry
A data object in a region. A region entry consists of a key and a value.
The value is either null (invalid) or an object. A region entry knows
what region it is in. See also region data, entry key, and entry value.
entry key
The unique identifier for an entry in a region.
entry value
The data contained in an entry.
expiration
A cached object expires when its time-to-live or idle timeout counters
are exhausted. A region has one set of expiration attributes for itself
and one set for all region entries.
expiration action
The action to be taken when a cached object expires. The expiration
action specifies whether the object is to be invalidated or destroyed,
and whether the action is to be performed only in the local cache or
throughout the distributed system. A destroyed object is completely
removed from the cache. A region is invalidated by invalidating all
entries contained in the region. An entry is invalidated by having its
value marked as invalid.
Expiration attributes are set at the region level for the region and at the
entry level for entries. See also idle timeout and time-to-live.
factory method
An interface for creating an object which at creation time can let its
subclasses decide which class to instantiate. The factory method helps
instantiate the appropriate subclass by creating the correct object from
a group of related classes.
278
Glossary
Pivotal GemFire Native Client
idle timeout
The amount of time a region or region entry may remain in the cache
unaccessed before being expired. Access to an entry includes any
get operation and any operation that resets the entry's time-to-live
counter. Region access includes any operation that resets an entry
idle timeout, and any operation that resets the region's time-to-live.
Idle timeout attributes are set at the region level for the region and
at the entry level for entries. See also time-to-live and expiration
action.
interest list
A mechanism that allows a region to maintain information about
receivers for a particular key-value pair in the region, and send out
updates only to those nodes. Interest lists are particularly useful when
you expect a large number of updates on a key as part of the entry life
cycle.
invalid
The state of an object when the cache holding it does not have the
current value of the object.
invalidate
Remove only the value of an entry in a cache, not the entry itself.
listener
An event handler. The listener registers its interest in one or more
events and is notified when the events occur.
load factor
A region attribute that sets initial parameters on the underlying
hashmap used for storing region entries.
local cache
The part of the distributed cache that is resident in the current process.
This term is used to differentiate the cache where a specific operation
is being performed from other caches in the distributed system. See
also remote cache.
local scope
Enables a region to hold a private data set that is not visible to other
caches. See also scope.
LRU
Least Recently Used. Refers to a region entry or entries most eligible
for eviction due to lack of interest by client applications.
279
Glossary
Pivotal GemFire Native Client
LRU entries limit
A region attribute that sets the maximum number of entries to hold in a
caching region. When the capacity of the caching region is exceeded,
LRU is used to evict entries.
membership
Applications and cache servers connect to a GemFire
distributed system by invoking the static function
DistributedSystem::connect. Through this connection, the
application gains access to the APIs for distributed data caches. When
a C++ or .NET application connects to a distributed system, it specifies
the system it is connecting to by indicating the communication protocol
and address to use to find other system members.
netSearch
The method used by GemFire to search remote caches for a data
entry that is not found in the local cache region. This operates only on
distributed regions.
overflows
An eviction option that causes the values of LRU entries to be moved
to disk when the region reaches capacity. See disk policy.
persistence manager
The persistence manager manages the memory-to-disk and disk-tomemory actions for LRU entries. See overflows.
region
A logical grouping of data within a cache. Regions are used to store
data entries (see entry). Each region has a set of attributes governing
activities such as expiration, distribution, data loading, events, and
eviction control.
region attributes
The class of attributes governing the creation, location, distribution,
and management of a region and its entries.
region data
All of the entries directly contained in the region.
region entry
See entry.
remote cache
Any part of the distributed cache that is resident in a process other
than the current one. If an application or cache server does not have
a data entry in the region in its local cache, it can do a netSearch in
an attempt to retrieve the entry from the region in a remote cache. See
also local cache.
280
Glossary
Pivotal GemFire Native Client
scope
Region attribute. Identifies whether a region keeps its entries private
or automatically sends entry value updates to remote caches
and incorporates updates received from remote caches. The
scope also identifies whether distribution operations must wait for
acknowledgement from other caches before continuing. See also
distributed scope and local scope.
serialization
The process of converting an object graph to a stream of bytes.
server
In a client-server topology, the server manages membership and
allows remote operations. The server maintains membership
information for its clients in the distributed system, along with
information about peer applications and other servers in the system.
See also cache server.
system member
A process that has established a connection to a distributed system.
time-to-live
The amount of time a region or region entry may remain in the cache
without being modified before being expired. Entry modification
includes creation, update, and removal. Region modification includes
creation, update, or removal of the region or any of its entries.
Time-to-live attributes are set at the region level for the region, and
at the entry level for entries. See also idle timeout and expiration
action.
XML
EXtensible Markup Language. An open standard for describing
data from the W3C, XML is a markup language similar to HTML.
Both are designed to describe and transform data, but where HTML
uses predefined tags, XML allows tags to be defined inside the XML
document itself. Using XML, virtually any data item can be identified.
The XML programmer creates and implements data-appropriate tags
whose syntax is defined in a DTD file.
281
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement