Oracle Real Application Clusters (RAC) 11g Release 2

Oracle Real Application Clusters (RAC) 11g Release 2
An Oracle White Paper
November 2010
Oracle Real Application Clusters (RAC)
11g Release 2
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
Introduction ....................................................................................... 1
What is Oracle Real Application Clusters? ........................................ 2
Oracle Real Application Clusters Architecture ................................... 2
Oracle Clusterware........................................................................ 3
Hardware Architecture................................................................... 4
Oracle RAC on Extended Distance Clusters.................................. 8
Benefits of Oracle Real Application Clusters...................................... 9
High Availability ............................................................................. 9
ScScalability ................................................................................ 10
Managing Your Oracle Real Application Clusters Database ............ 11
Oracle Enterprise Manager.......................................................... 11
Rolling Patch Application ............................................................. 12
Rolling Release Upgrade Support ............................................... 13
Workload Management with Oracle Real Application Clusters ........ 14
Server Pools................................................................................ 14
Services ...................................................................................... 15
Connection Load Balancing......................................................... 15
Fast Application Notification ........................................................ 16
Load Balancing Advisory ............................................................. 16
Conclusion ...................................................................................... 17
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
Introduction
Oracle Real Application Clusters (RAC) allows Oracle Database to run any packaged or
custom application, unchanged across a server pool. This provides the highest levels of
availability and the most flexible scalability. If a server in the pool fails, the Oracle
database continues to run on the remaining servers. When you need more processing
power, simply add another server to the pool without taking users offline. To keep costs
low, even the highest-end systems can be built out of standardized, commodity parts.
Oracle Real Application Clusters provides a foundation for Oracle’s Private Cloud
Architecture. Oracle RAC technology enables a low-cost hardware platform to deliver the
highest quality of service that rivals and exceeds the levels of availability and scalability
achieved by more expensive mainframe SMP computers. By dramatically reducing
administration costs and providing new levels of administration flexibility, Oracle RAC
enables private clouds. Oracle RAC 11g Release 2 in addition enables customers to
build a dynamic private cloud infrastructure.
This paper provides a technical overview of Oracle Real Application Clusters 11g with the
emphasis on the features and functionality that can be implemented to provide the
highest availability and scalability for enterprise applications.
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
“We’ve been able to save over $5 million dollars a year by re-platforming
from our mainframe to Oracle Real Application Clusters.”
— Eugene Park, Senior Director of Platform Services, PG&E
What is Oracle Real Application Clusters?
Oracle Real Application Clusters is an option to the Oracle Database Enterprise Edition that was
first introduced with Oracle 9i. Oracle Real Application Clusters is now proven technology used
by thousands of customers in every industry for every type of application. Oracle RAC provides
options for scaling applications beyond the capabilities of a single server. This allows customers to
take advantage of lower cost commodity hardware to reduce their total cost of ownership and
provide a scaleable computing environment that supports their application workload. Oracle RAC
enables the Oracle Database to run mainstream business applications of all kinds on server pools
including popular packaged products (such as Oracle Applications, Peoplesoft, SAP), in-house
developed applications, which can be either OLTP, DSS, or a mixed workload.
Oracle Real Application Clusters is a key component of Oracle’s Maximum Availability
Architecture (MAA)1, which provides direction to architect the highest availability for applications.
Oracle RAC provides the ability to remove the server as a single point of failure in any database
application environment.
Oracle Real Application Clusters Architecture
An Oracle RAC database is a clustered database. A cluster can be described as a pool of
independent servers that cooperate as a single system. Server Pools provide improved fault
resilience and modular incremental system growth over single symmetric multi-processor (SMP)
systems. In the event of a system failure, clustering ensures high availability to users. Access to
mission critical data is not lost. Redundant hardware components, such as additional servers,
network connections, and disks, allow the cluster to provide high availability. Such redundant
hardware architectures avoid a single point-of-failure and provide exceptional fault resilience.
1 For more information on Oracle’s Maximum Availability Architecture see:
http://www.oracle.com/technetwork/database/features/availability/maa-090890.html
-2-
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
Figure 1 Oracle Real Application Clusters Architecture
With Oracle Real Application Clusters, Oracle de-couples the Oracle Instance (the processes and
memory structures running on a server to allow access to the data) from the Oracle database (the
physical structures residing on the storage storing the data, commonly referred to as the datafiles).
A clustered database is a single database that can be accessed by multiple instances. Each instance
runs on a separate server in the server pool. When additional resources are required, additional
servers and instances can easily be added to the server pool with no downtime. Once the new
instance is started, applications using services can immediately take advantage of it with no changes
to the application or application server.
Oracle Real Application Clusters is an extension of the Oracle Database and therefore benefits
from the manageability, reliability and security features built into Oracle Database 11g.
Oracle Clusterware
Starting with Oracle Database 10g, Oracle provides Oracle Clusterware, a portable clusterware
solution that is integrated and designed specifically for the Oracle Database. Oracle Clusterware
provides a complete clustering solution and supports any application. Oracle Clusterware is a prerequisite for all Oracle RAC implementations. Support is made easier as there is one support
organization to deal with for the clusterware and cluster database. One can choose to run Oracle
-3-
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
RAC on top of additional, selected third party cluster solutions, in which case, Oracle Clusterware
must still manage all Oracle RAC databases on the system.
Oracle Clusterware monitors and manages Oracle Real Application Cluster databases. When a
server in the server pool is started, all instances, listeners and services are automatically started. If
an instance fails, the Oracle Clusterware will automatically restart the instance so that the service is
often restored before the administrator notices it was down.
With Oracle Database 10g Release 2, Oracle added a High Availability API so that non-Oracle
applications can be managed by the high availability framework provided by Oracle Clusterware.
When registering the application with Oracle Clusterware, information must be provided on how
to start, stop, and monitor the process. In addition, one can specify candidate servers to take over a
resource once a failure occurred. With Oracle Database 11g Release 2, managing applications was
made even easier through the graphical interface provided by Oracle Enterprise Manager (EM).
Using Oracle EM and the enhanced High Availability Framework provided by Oracle Clusterware
in conjunction with even more dependency options, Oracle Clusterware can now be use to easily
model even complex failover and recovery scenarios.
Figure 2: Oracle Clusterware Overview
Hardware Architecture
Oracle Real Application Clusters is a shared everything architecture. All servers in the server pool
share all storage used for an Oracle RAC database. The type of storage pool used can be network
attached storage (NAS), storage area network (SAN), or SCSI disks. Your storage choice is dictated
by the server hardware choice and the hardware supported by your hardware vendor. The key to
choosing an appropriate storage pool is choosing a storage system that will provide scaleable I/O
for your application and an I/O system that will scale as additional servers are added to the pool.
-4-
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
An Oracle RAC Database requires a network connection to the Local Area Network (LAN) that a
database server is attached to for application connections. A server pool also requires a private
network commonly known as “the interconnect”. Oracle recommends redundant interconnects for
high availability purposes. With Oracle Real Application Clusters 11g Release 2, patch set 1
(11.2.0.2), Oracle provides a native solution to support redundant interconnects as past of the
Oracle Grid Infrastructure solution. With former versions, an external, mostly Operating System
(OS) dependent network redundancy solution (bonding / teaming) should be used to provide
failover and load balancing. The interconnect is used by Oracle Clusterware for inter-node
messaging. The interconnect is also used by Oracle RAC to implement the cache fusion technology.
Oracle recommends the use of UDP over GigE for the cluster interconnect. The use of crossover
cables as the interconnect is not supported for a production Oracle RAC database.
A server pool is made up of 1 or more servers, each having a public LAN connection, an
interconnect connection, and must be connected to a shared pool of storage. Oracle Clusterware
and Oracle Real Application Clusters support up to 100 nodes in the cluster. Each server in the
cluster does not have to be exactly the same but it must run the same operating system, and the
same version of Oracle. All servers must support the same architecture; e.g. all servers must be
either 32bit or 64bit.
Current and detailed information on certifications and technology restrictions related to Oracle Real
Application Clusters can be obtained through My Oracle Support (http://support.oracle.com)
File Systems and Volume Management
Since Oracle RAC is a shared everything architecture, the volume management and file system used
must be cluster-aware. Oracle recommends using Oracle Automatic Storage Management (ASM),
which is a feature included with Oracle Database 11g to automate the management of various storage
pools for the database. ASM provides the performance of an async I/O storage subsystem with the
ease of management of a file system. ASM distributes I/O load across all available resource to
optimize performance while removing the need for manual I/O tuning. With Oracle Database 11g
Release 2 ASM comes with a dynamic volume manager and a general purpose file system.
Alternatively, Oracle supports certain, certified cluster file systems such as the Oracle Cluster File
System (OCFS) that is available on Windows and Linux (called OCFS2). With Oracle Database 11g
Release 2, the Oracle Universal Installer and the Database Configuration Assistant do not support the
use of raw or block devices for database files anymore. Databases that are currently using raw devices
can continue to use those during or after an upgrade to 11g Release 2 as the command line interfaces
continue to support the use of raw devices.
-5-
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
Oracle Grid Infrastructure
With Oracle RAC 11g Release 2, Oracle introduces Oracle Grid Infrastructure. With Oracle Grid
Infrastructure, Oracle integrated Oracle ASM, the proven storage pool management solution for the
Oracle Database, and Oracle Clusterware in one software bundle. Oracle has thereby combined two
of its strongest products for cluster environments to form a universal grid foundation. In an Oracle
RAC environment, Oracle Grid Infrastructure provides the necessary foundation with respect to
volume management, file system, and server pool management to run an Oracle RAC database. The
system administrator and systems administration team should therefore manage oracle Grid
Infrastructure.
Virtual Internet Protocol Address (VIP)
Oracle Real Application Clusters 11g requires a virtual IP address for each server in the cluster. The
virtual IP address is an IP address managed by Oracle Clusterware and taken from the same subnet
as the Local Area Network (LAN). This address is used by applications to connect to the Oracle
RAC database. If a node fails, the Virtual IP is failed over to another node in the cluster to provide
an immediate “node down”-response to incoming connection requests. This increases the
availability for applications, as they no longer have to wait for network timeouts before the
connection request fails over to another instance in the server pool.
Grid Naming Service
Oracle RAC 11g Release 2 introduces the Grid Naming Service (GNS), which makes it easier to scale
by automating the VIP management for Oracle RAC. In order to use GNS, one will have to work
with the Network Administrator to set up a delegated domain in the Domain Name Service (DNS)
and a Virtual IP address for the Grid Naming Service (GNS). However, once this part is set up,
Oracle will manage the VIPs in the cluster automatically. Regardless of whether or not you add or
remove servers from the pool, you will not have to return to the network administrator for additional
IP addresses. To further automate the network management in the cluster, the use of the Grid
Naming Service (GNS) requires a DHCP server on the public network. Oracle uses DHCP to
dynamically allocate the required Virtual IP addresses as servers join the cluster. Overall, GNS will
reduce unnecessary, manual re-configuration with respect to the network in dynamic environments.
Single Client Access Name (SCAN)
Oracle RAC 11g Release 2 also introduces the new Single Client Access Name (SCAN) to simplify
client access to Oracle RAC databases. SCAN provides a single name to be used in the client
connection requests that does not change as the cluster expands or if any of the nodes in the cluster
changes over time. This allows the use of simplified connect strings such as EZConnect (sqlplus
system/[email protected]:1521/oltp - or for JDBC: jdbc:oracle:thin:@sales1-scan:1521/oltp).
When using GNS, you only need to supply the name and listener port for your SCAN. If you choose
not to use GNS, you must define the SCAN in your DNS as a single name that round robins to 3 IP
Addresses. These IP addresses must be on the same subnet as the public network for the cluster.
-6-
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
Cluster Verification Utility
Oracle Grid Infrastructure includes a cluster configuration verification tool (CVU). The cluster
verification tool eliminates errors through pre- and post-validation of installation steps and/or
configuration changes. It can also be used for ongoing cluster validation, for which reason the toll has
been incorporated into Oracle Grid Infrastructure 11g Release 2, patch set 1 (11.2.0.2) The tool can
invoked through a command line interface. With Oracle RAC 11g Release 2, the Cluster Verification
Utility (CVU), is integrated with the OUI to validate configuration input during the installation
interview as well as the system pre-requisites for Oracle Clusterware, ASM, and Oracle Database. If
your server(s) do not meet certain necessary pre-requisites, a fix-up script is created by CVU. In order
to automatically fix those pre-requisites, you can then run the fix-up script on each node that applies.
Figure 3: Oracle Universal Installer CVU Integration
-7-
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
“High availability is absolutely essential for us…we now use Oracle RAC for instance failover, data guard for site failover,
ASM to manage our storage, and Oracle clusterware to hang the whole thing together.”
— Jon Waldron, Executive Architect, Commonwealth Bank of Australia
Oracle RAC on Extended Distance Clusters
Oracle RAC on Extended Distance Clusters is an architecture where servers in the cluster reside in
locations that are physically separate. Oracle RAC on Extended Distance Clusters provides extremely
fast recovery from a site failure and allows for all servers, in all sites, to actively process transactions as
part of a single database cluster. While this architecture creates great interest and has been successfully
implemented, it is critical to understand where this architecture best fits especially in regards to distance,
latency, and degree of protection it provides.
The high impact of latency, and therefore distance, creates some practical limitations as to where this
architecture can be deployed. This architecture fits best where the 2 datacenters are located relatively
close (<~100km) and where the extremely expensive costs of setting up direct cables with dedicated
channels between the sites have already been taken.
Oracle RAC on Extended Distance Clusters provides greater availability than local Oracle RAC but it
may not fit the full Disaster Recovery requirements of every organization. Feasible separation is great
protection for some disasters (local power outage, airplane crash, server room flooding) but not all.
Disasters such as earthquakes, hurricanes, and regional floods may affect a greater area. One should
analyze their situation in order to determine if both sites are likely to be affected by the same disaster.
For comprehensive protection against disasters including protection against corruptions and regional
disasters, Oracle recommends the use of Oracle Data Guard in combination with Oracle RAC as
described in the Oracle Maximum Availability Architecture guidelines. Oracle Data Guard also provides
additional benefits such as support for rolling upgrades across Oracle versions.
Configuring an extended distance cluster is more complex than configuring a local cluster. Specific
focus needs to go into node layout, voting disks, and data disk placement. Implemented properly, this
architecture can provide greater availability than a local Oracle RAC database, but should be considered
with Oracle’s Maximum Availability Architecture in mind.
-8-
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
Benefits of Oracle Real Application Clusters
High Availability
Oracle Real Application Clusters 11g provides the foundation for data centre-high availability. It is also
an integral component of Oracle’s Maximum Availability Architecture, which provides best practices to
provide the highest availability for your data center. Oracle Real Application provides the following ke
characteristics essential for a high available data management::
Reliability – The Oracle Database is known for its reliability. Oracle Real Application Clusters takes this
a step further by removing the database server as a single point of failure. If an instance fails, the
remaining instances in the server pool remain open and active. Oracle Clusterware monitors all Oracle
processes and immediately restarts any failed component.
Recoverability – The Oracle Database includes many features that make it easy to recover from all
types of failures. If an instance fails in an Oracle RAC database, it is recognized by another instance in
the server pool and recovery will start automatically. Fast Application Notification (FAN) and Fast
Connection Failover (FCF) or Transparent Application Failover (TAF) make it easy for applications to
mask component failures from the user.
Error Detection – Oracle Clusterware automatically monitors Oracle RAC databases as well as other
Oracle processes (ASM, listener, etc) and provides fast detection of problems in the environment. It also
automatically recovers from failures often before users noticed that a failure has occurred. Fast
Application Notification (FAN) provides the ability for applications to receive immediate notification of
cluster component failures in order to re-issue the transaction before the failure surfaces.
Continuous Operations – Oracle Real Application Clusters provides continuous service for both
planned and unplanned outages. If a server (or an instance) fails, the database remains open and the
application is able to access data. Most database maintenance operations can be completed without
downtime and are transparent to the user. Many other maintenance tasks can be done in a rolling
fashion so application downtime is minimized or removed. Fast Application Notification and Fast
Connection Failover assist applications in meeting service levels.
-9-
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
“We successfully completed a ‘Big Bang’ migration from a mainframe system to a clustered server
environment. The Oracle environment is highly stable and offers powerful performance and easy scalability,
which has enhanced user satisfaction and customer service.”
— NoCheol Park, CIO & Senior VP, NGM Program Manager, SK Telecom
Scalability
Oracle Real Application Clusters provides a unique technology for scaling applications. Traditionally,
when database servers ran out of capacity, they were replaced with new and larger servers. As servers
grow in capacity, they are more expensive. For databases using Oracle RAC, there are alternatives for
increasing the capacity. Applications that have traditionally run on large SMP servers can be migrated to
run on pools of small servers. Alternatively, you can maintain the investment in the current hardware and
add a new servers to the pool (or to create a server pool) to increase the capacity. Adding servers to a
server pool with Oracle Clusterware and Oracle RAC does not require an outage and as soon as the new
instances are started, the application can take advantage of the extra capacity. All servers in the server pool
must run the same operating system and the same version of Oracle, but they do not have to be of exactly
the same capacity. Customers today run server pools that fit their needs often using servers of (slightly)
different characteristics.
The Oracle Real Application Clusters architecture automatically accommodates rapidly changing business
requirements and the resulting workload changes. Application users, or mid tier application server clients,
connect to the database by way of a service name. Oracle automatically balances the user load among the
multiple nodes in the server pool. The Oracle Real Application Clusters database instances on the
different nodes subscribe to all or some subset of database services. This provides the Database
Administrator witjh the flexibility of choosing whether specific application clients that connect to a
particular database service can connect to some or all of the database nodes. Administrators can
painlessly add processing capacity as application requirements grow. The Cache Fusion architecture of
Oracle RAC immediately utilizes the CPU and memory resources of the new node(s). DBAs do not need
to manually re-partition the data.
Another way of distributing workload in an Oracle database is through the Oracle Database parallel
execution feature. Parallel execution (I.E. parallel query or parallel DML) divides the work of executing a
SQL statement across multiple processes. Using Oracle’s cost-based optimizer intelligent decisions are
made with respect to intra-node and inter-node parallelism in an Oracle RAC environment.
For example, if a particular query requires six query processes to complete the work and six CPUs are idle
on the local node (the node that the user connected to), then the query is processed using only local
resources. This demonstrates efficient intra-node parallelism and eliminates the query coordination
overhead across multiple nodes. However, if there are only two CPUs available on the local node, then
those two CPUs and four CPUs of another node are used to process the query. In this manner, both
inter-node and intra-node parallelism are used to provide speed up for query operations.
- 10 -
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
Managing Your Oracle Real Application Clusters Database
Oracle Real Application Clusters provides a single system image for easy configuration and
management. The Oracle RAC database can be installed, configured, and managed from a single
location. All tools and utilities provided to manage the database are cluster-aware from the Oracle
Universal Installer (OUI), to Enterprise Manager including the database configuration assistant
(DBCA), the database upgrade assistant (DBUA), the network configuration assistant (NETCA), and
the command line interfaces such as srvctl.
Oracle Enterprise Manager
Oracle Enterprise Manager Grid Control is the recommended management interface for an Oracle
environment. Oracle Grid Control delivers centralized management functionality for the complete
Oracle IT infrastructure, including systems running Oracle and non-Oracle technologies. With a broad
set of administration, configuration management, provisioning, end-to-end monitoring, and security
capabilities, Oracle Grid Control reduces the cost and complexity of managing private clouds, while
helping customers maintain their IT infrastructure service levels.
Oracle Enterprise Manager Database Control is the graphical management tool provided by Oracle to
manage your Oracle Database. Oracle Enterprise Manager Database Control can be automatically
configured by the DBCA when a database is created. Alternatively, the database can be automatically
registered with Oracle Enterprise Manager Grid Control at the time of creation. Both Enterprise
Manger products are cluster-aware and can be used to manage Oracle RAC and server pools created
with Oracle Clusterware.
Figure 4: Oracle Enterprise Manager Database Control - Server Pool Management
- 11 -
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
Oracle Enterprise Manager Grid Control simplifies the day to day management tasks to be performed by
an Oracle Database Administrator as well as Oracle Database deployments and monitoring. It assists:
•
When migrating Oracle Databases from a file system to Oracle ASM
•
When converting an Oracle Single Instance into an Oracle RAC Database
•
When monitoring various HA areas (e.g. backup & recovery, replication, disaster recovery)
•
When defined thresholds are reached (cluster-wide) so that an alert must be issued.
Starting with Oracle Enterprise Manager 10g Release 2, Grid Control provides additional capabilities to
make the provisioning of Oracle Real Application Clusters databases easier. The initial creation of a
server pool, including deploying the required Oracle software (a.k.a. Oracle Homes) and the
configuration of Oracle clusterware can easily be performed using Oracle Enterprise Manager.
The Oracle Home-software can be kept in Oracle Enterprise Manager as the known “Gold Image” or
sourced for further deployments. The “Gold Image” is created from a copy of a known, good
implementation of Oracle Clusterware or Oracle Real Application Clusters. Thereby, Oracle Enterprise
Manager supports the complete end-to-end creation of new Oracle RAC and Oracle Grid Infrastructure
environments, including the execution of super-user actions (root.sh) and customizable pre- and postinstallation steps. Similar support is provided when extending the server pools with more servers.
For Linux operating systems, Oracle can also provision an “image” to a bare metal node. The image can
consist of the Operating System, the Oracle Enterprise Manager agent, Oracle Grid Infrastructure, and
Oracle Database with Oracle Real Application Clusters. This image can be associated with a hardware
profile. All the components for this image are stored as "Gold Images" in Oracle Enterprise Manager.
Oracle Enterprise Manager Database Control introduces a new area for server pool and cluster
management. The “cluster-tab” allows you to monitor and manage a server pool created with Oracle
Clusterware. From the cluster-tab, you can create and manage resources for both Oracle as well as user
created resources. Users can create resources with dependencies to allow the cluster to monitor and
managed any process on any server in the cluster. Oracle Clusterware includes sophisticated dependency
options to allow you to easily model all your business critical applications. Oracle Enterprise Manager
Database Control 11g Release 2 requires an Oracle RAC 11g Release 2 database. However, it can manage
Oracle Clusterware and its resources even when the database is down.
Rolling Patch Application
Oracle supports the application of patches to the nodes of an Oracle RAC database in a rolling fashion
with no downtime. Patches are applied one node at a time while the other nodes in the Oracle RAC
system are up and operational. This requires that each node has a separate Oracle Home. Patches will be
labeled as being qualified for installation as rolling upgradeable, or not, depending on the changes being
made by the patch. Some patches that modify common structures shared between instances, or the
contents of the database, will not be.
- 12 -
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
This capability is supported beginning with Oracle 9.2.0.2. All Oracle Clusterware patches can be applied
in a rolling fashion. With Oracle Database 11g, Automatic Storage Management is rolling upgradeable.
Oracle Grid Infrastructure 11g Release 2, which includes Oracle Clusterware and Oracle ASM, is rolling
upgradeable for all sorts of patches, including patch sets. Also, with Oracle Database 11g Release 2,
Oracle has streamlined the application of patches in a cluster. The patch utility Opatch, will apply
patchsets and patch bundles in a few easy steps.
Rolling Release Upgrade Support
Oracle Grid Infrastructure 11g Release 2 supports rolling upgrades; so does Oracle Clusterware, which
has offered this feature starting with Oracle Clusterware 10g. This provides the ability to upgrade without
taking the entire cluster down and therefore enables 24x7 operation of business. Once you have upgraded
Oracle Automatic Storage Management to 11g, you will be able to do rolling upgrades with ASM, too.
Oracle RAC 11g supports database software upgrades (from Oracle Database 10g Release 1 Patchset 1
onwards) in a rolling fashion – with nearly zero database downtime, by using Oracle Data Guard SQL
Apply. The steps involve upgrading the logical standby database to the next release, running in a mixed
mode to test and validate the upgrade, doing a role reversal by switching over to the upgraded database,
and then finally upgrading the old primary database. While running in a mixed mode for testing purpose,
the upgrade can be aborted and the software downgraded, without data loss. For additional data
protection during these steps, a second standby database may be used.
By supporting rolling upgrades with minimal downtimes, Oracle Data Guard reduces the large
maintenance windows typical for many administrative tasks, and enables 24x7 operation of business.
- 13 -
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
Workload Management with Oracle Real Application Clusters
Applications using an Oracle RAC database need to manage workload across a server pool. Oracle Real
Application Clusters includes innovative technologies to manage workloads while providing the best
application throughput given the configuration and high availability for the application. Oracle RAC 11g
Release 2 introduces a simplified management of Oracle RAC databases within server pools to ease the
consolidation of applications into a pool of servers, while maintaining resource allocation and role
separated management at the same time.
App
Servers
ERP
CRM
FREE
DW
RAC ONE
NODE
Figure 5 Consolidate into Low Cost Servers using Server Pools
Server Pools
With Oracle RAC 11g Release 2, a database can be defined to run in a server pool. A server pool is a
logical entity in the cluster that allows the administrator to allocate resources to specific applications. A
server pool is defined by 3 attributes: min (the minimum number of servers in the pool, max (the
maximum number of servers in the pool), and importance (providing the ability to give a relative
importance to different pools in the cluster). Oracle Clusterware will allocate servers to user defined pools
when a cluster reconfiguration takes place. Oracle Clusterware will allocate servers in order of the
importance. The number of instances maintained for the database is defined by the cardinality of the
server pool.
- 14 -
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
The database configuration assistant (DBCA) provides the option to create a policy-managed database by
defining the cardinality and the name. A server pool will automatically be created trying to allocate as
many servers and therefore instances of the database as specified by the cardinality parameters requested.
Oracle Clusterware will maintain the servers in the server pool up to the maximum based on availability of
servers in the cluster.
Services
Workload Management relies on the use of Services, a feature of the Oracle Database. Services hide the
complexity of an Oracle RAC database by providing a single system image to manage workload. Services
allow applications to benefit from the reliability of a server pool. Traditionally, a database provided a
single service and this name was used in the connect data for SQL*NET. With Oracle Database 11g, a
DBA can define more than a hundred database services for a single database.
This allows you to breakup workloads from applications into manageable components based on business
requirements such as service levels and priorities. Services are integrated with many features of the Oracle
Database. Application users can – for example – automatically be assigned to a Resource Manager
consumer group, which can limits their consumed resources such as CPU, depending on a service. Batch
jobs can be assigned to specific job class based on their service. The use of services also achieves location
transparency for queues when using Oracle Streams Advanced Queuing, while inter-node parallel query
will be restricted to the instances where the service is active with Oracle Database 11g.
A service can span one or more instances of an Oracle database and an instance can support multiple
services. The number of instances offering a service is managed dynamically by the DBA independently of
the application. When outages occur, services are automatically restored to surviving instances. When
instances are restored, any services that are not running are restored automatically.
With Policy Managed Databases, a service can only run in one server pool and is defined as either
uniform (service is offered by all instances in the server pool) or as a singleton (runs only on one instance
in the server pool).
Connection Load Balancing
Oracle Net Services provide connection load balancing for database connections. Client-side load
balancing, which balances connection requests across all SCAN listeners in the cluste, is achieved by using
the SCAN on the address list of the client connect string. SQL*NET will randomly select one of the
SCAN ip addresses. If the server chosen is not available, the next server in the list is tried. Server side load
balancing is achieved using the SCAN listener. Each SCAN listener is aware of all instances in the cluster
providing each service. Based on goal defined for the service, the listener chooses the instance that will
best meet the goal and the connection is routed to that instance through the local listener.
- 15 -
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
"The clustered database architecture that we've deployed has satisfied our business and performance requirements. It also offers us
the flexibility we need for future growth. The newly re-architected Global Data Warehouse features, in addition to more powerful
processors, improved performance and provide additional capacity for applications and users."
— Matthew Schroeder, Manager, Business Information & Technologies, Alcoa
Fast Application Notification
Fast Application Notification (FAN) provides integration between the Oracle RAC database and the
application. It allows the application to be aware of the current configuration of the server pools at any
given time so that application connections are only made to those instances that are currently able to
respond to the application requests. The Oracle RAC HA framework posts a FAN event immediately
when a state change occurs within the cluster.
Integrated clients receive these events and immediately react. For DOWN events, application
interruption is minimized by cleaning up connections to the failed instance, in-flight transactions are
interrupted with an error returned to the application. Applications making connections are directed to
active instances only. Server side callouts can be used to log trouble tickets or page administrators alerting
them of the failure.
For UP events, new connections are created to allow the application to immediately take advantage of the
extra resources available. Oracle JDBC, the Oracle Universal Connection Pool, ODP.NET and OCI
clients are integrated with FAN. Applications with their own connection pools can take advantage of
FAN by using either the Oracle RAC FAN API available with the Oracle Database 11g Release 2 JDBC
driver or the Oracle Call Interface callback functionality.
Load Balancing Advisory
Since Database workloads can change over time as well as the server pool configuration, it is important to
create and allocate database connections based on the most current information. Oracle Real Application
Clusters provides a load balancing advisory. Oracle RAC constantly monitors the workload being
executed for each service by each instance providing the service. This information is published to the
Automatic Workload Repository and published to the application using FAN events. The FAN event
includes the current service level provided and a recommendation of what percentage of connections to
be directed to each instance.
The integrated Oracle Clients use these events to provide intelligent load balancing of application
requests. Most connection pools use a random or round robin algorithm to select an idle connection from
the pool when the application attempts to connect. Using FAN events from the load balancing advisory,
the connection pool will select the connection currently providing the best service. Oracle JDBC, Oracle
UCP, OCI, and ODP.NET provide runtime connection load balancing through integration with the load
balancing advisory.
- 16 -
Oracle White Paper—Oracle Real Application Clusters 11g Release 2
Figure 6: Load Balancing in an Oracle RAC stack
Conclusion
Oracle Real Application Clusters has been designed for high availability and scalability. By providing
protection from hardware and software failures, Oracle Real Application Clusters provides system
availability ensuring continuous data access. Its scale out and scale up features offer a platform, which can
grow in any direction allowing enterprises to grow their businesses. Existing applications as well as newly
developed applications benefit from the transparency Oracle Real Application Clusters provides.
Application development as well as administration and change management thus become much easier,
allowing reduction in total cost of ownership. Oracle Real Application Clusters is unique to the market
with its offering and capabilities. Oracle RAC is used by thousands of customers worldwide in all
industries in mission critical and many other application environments and a crucial component when
setting up or managing private clouds.
- 17 -
Oracle Real Application Clusters (RAC)
11g Release 2
November 2010
Author: Barb Lundhild, Markus Michalewicz
Contributing Authors:
Oracle Corporation
World Headquarters
500 Oracle Parkway
Redwood Shores, CA 94065
U.S.A.
Copyright © 2010, Oracle and/or its affiliates. All rights reserved. This document is provided for information purposes only and
the contents hereof are subject to change without notice. This document is not warranted to be error-free, nor subject to any other
warranties or conditions, whether expressed orally or implied in law, including implied warranties and conditions of merchantability or
fitness for a particular purpose. We specifically disclaim any liability with respect to this document and no contractual obligations are
formed either directly or indirectly by this document. This document may not be reproduced or transmitted in any form or by any
means, electronic or mechanical, for any purpose, without our prior written permission.
Worldwide Inquiries:
Oracle is a registered trademark of Oracle Corporation and/or its affiliates. Other names may be trademarks of their respective
Phone: +1.650.506.7000
owners.
Fax: +1.650.506.7200
oracle.com
0109
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