Setting up Oracle 11g Data Guard for SAP customers

Setting up Oracle 11g Data Guard for SAP customers
An Oracle White Paper
January 2010
Setting up Oracle 11g Data Guard
for SAP customers
Database Version: 11g Release 2
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Preface ................................................................................................ 1
Requirements ...................................................................................... 6
Aim of the document ........................................................................... 6
Preparatory work ................................................................................. 7
Structure of the standby database .................................................... 14
Creating standby redo log files .......................................................... 17
Configuring Data Guard Broker ......................................................... 18
Deploying the transfer of logs and function check ............................ 24
Implementing SAP BRARCHIVE with Data Guard ........................... 25
Disaster handling / role transition ...................................................... 27
Reconnect SAP instance to database ............................................... 46
Performance ...................................................................................... 50
Annex ................................................................................................ 52
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Preface
The document explains the structure of a physical standby database with Oracle Data
Guard in an SAP environment. It indicates all the steps needed to successfully install and
configure an Oracle Data Guard system with a physical standby database and the logical
order in which they must be carried out.
To enable you to operate the standby database (Oracle Data Guard), a description of
how to configure the Data Guard Broker is also provided. In just a few steps this service
allows you to swap the database roles. This means that in the event of a disaster, what is
known as a switchover or failover is undertaken almost automatically. The database
administrator can initiate the process with just one command.
Overview of Oracle Data Guard
Oracle Data Guard is the management, monitoring, and automation software
infrastructure that creates, maintains, and monitors one or more standby databases to
protect enterprise data from failures, disasters, errors, and data corruptions.
Data Guard maintains standby databases as consistent copies of the production
database as far as transactions are concerned. These standby databases can be located
at remote disaster recovery sites thousands of miles away from the production data
center, or they may be located in the same city, same campus, or even in the same
building. If the production database becomes unavailable because of a planned or an
unplanned outage, Data Guard can switch any standby database to the production role,
thus minimizing the downtime associated with the outage, and preventing any data loss.
1
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Data Guard functionality
A Data Guard configuration consists of a production database, also known as the primary
database, and up to nine standby database(s), which are consistent copies of the primary database
as far as transactions are concerned. Data Guard maintains this transactional consistency using
redo data. As transactions occur in the primary database, redo data is generated and written to
the local redo log files. With Data Guard, this redo data is also transferred to the standby sites
and applied to the standby databases, keeping them synchronized with the primary database.
Data Guard allows the administrator to choose whether this redo data is sent synchronously or
asynchronously to a standby site.
The underlying technologies for standby databases are Data Guard Redo Apply (physical standby
database), and Data Guard SQL Apply (logical standby database). A physical standby database
has on-disk database structures that are identical to the primary database on a block-for-block
basis, and is updated using Oracle media recovery. A logical standby database is an independent
database that contains the same data as the primary database. It is updated using SQL statements,
and has the advantage that it can be used for recovery and for other tasks such as reporting and
queries at the same time.
Data Guard enables role transitions between the primary database and a chosen standby
database, reducing overall downtime during planned outages and unplanned failures.
2
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
The primary and standby databases, as well as their various interactions, may be managed using SQL*Plus.
Data Guard also offers a distributed management framework called the Data Guard Broker, which automates
and centralizes the creation, maintenance, and monitoring of a Data Guard configuration. For easier
manageability, administrators may use either Oracle Enterprise Manager or the Broker’s own specialized
command-line interface (DGMGRL) to take advantage of the Broker’s management capabilities.
The following diagram shows the Data Guard components.
Benefits of Data Guard
•
Disaster recovery and high availability - Data Guard provides an efficient and comprehensive disaster recovery
and high availability solution. Automatic failover and easy-to-manage switchover capabilities allow quick
role transitions between primary and standby databases, minimizing the downtime of the primary
database for planned and unplanned outages.
•
Complete data protection - A standby database also provides an effective safeguard against data corruptions
and user errors. Storage level physical corruptions on the primary database do not spread to the standby
database. Similarly, logical corruptions or user errors that cause the primary database to be permanently
damaged can be resolved. Finally, the redo data is validated at the time it is received at the standby
database and also when applied to the standby database.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
•
Efficient utilization of system resources - A physical standby database can be used for backups and read-only
reporting, thereby reducing the primary database workload and saving valuable CPU and I/O cycles. In
Oracle Database 10g Release 2, a physical standby database can also be easily converted back and forth
between being a physical standby database and an open read/write database. A logical standby database
allows its tables to be simultaneously available for read-only access while they are updated from the
primary database. A logical standby database also allows users to perform data manipulation operations
on tables that are not updated from the primary database. Finally, additional indexes and materialized
views can be created in the logical standby database for better reporting performance.
•
Flexibility in data protection to balance availability against performance requirements - Data Guard offers the
Maximum Protection, Maximum Availability and Maximum Performance modes to help enterprises
balance data protection against system performance requirements.
•
Protection from communication failures - If network connectivity is lost between the primary and one or more
standby databases, redo data cannot be sent from the primary database to those standby databases
affected. Once connectivity is reestablished, the missing redo data is automatically detected by Data
Guard and the necessary archive logs are automatically transmitted to the standby databases. The standby
databases are resynchronized with the primary database, with no manual intervention by the
administrator.
•
Centralized and simple management - Data Guard Broker automates management and monitoring tasks across
the multiple databases in a Data Guard configuration. Administrators may use either Oracle Enterprise
Manager or the Broker’s own specialized command-line interface (DGMGRL) to take advantage of this
integrated management framework.
•
Integrated with Oracle database - Data Guard is available as an integrated feature of the Oracle Database
(Enterprise Edition) at no extra cost.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Data Guard process architecture
As shown in the following figure, Data Guard uses several processes of the Oracle database instance to
achieve the automation necessary for disaster recovery and high availability.
On the primary database, Data Guard uses the Log Writer (LGWR) process or multiple Archiver (ARCH)
processes to collect transaction redo data. In order to ensure isolation from network disruptions, the Log
Writer process uses specialized background processes, called LogWriter Network Server (LNS) process, to
synchronously or asynchronously transmit the redo data to the standby database. The Archiver processes
transmit the redo data to the standby database directly. The primary database also has the Fetch Archive Log
(FAL) process to provide a client-server mechanism for transmitting archived logs to the standby database
following a loss of communication between the primary and standby database(s), for automatic gap resolution
and resynchronization.
On the standby database, Data Guard uses one or more Remote File Server (RFS) processes to receive redo
data from the primary database, the Managed Recovery Process (MRP) to apply redo data to the physical
standby database, and the Logical Standby Process (LSP) to apply SQL-translated redo data to the logical
standby database.
If the Data Guard Broker is enabled, Data Guard also uses the Data Guard Broker Monitor (DMON) process
to manage and monitor the primary and standby databases as a unified configuration.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Requirements
A second database host, configured in exactly the same way as the first, is needed to operate Oracle Data
Guard. This means:
•
Same operating system, e.g. AIX 5.3 ML6 on both hosts
•
Same parameter settings on database and operating system (e.g. nfiles)
•
Identical Oracle version, currently the 11.2.0.1 database patch set with individual patches as
recommended in SAP note.
•
Identical file system structure, especially for SAP data and Oracle home.
•
The databases must be operated in ARCHIVELOG mode
•
Use of server parameter file (SPFILE)
Please read SAP note 105047 on approving Oracle Data Guard for an SAP environment.
Aim of the document
The document provides step-by-step instructions for how to set up an Oracle Data Guard solution with a
physical standby database. This should enable the user to manage the Data Guard system created using the
Data Guard Broker.
z
Learn how to setup a Data Guard configuration and the needed environment. There are prepared
examples for setup all the configuration files and sql scripts at the annex.
z
Read about the possibilities of that solution in case of a desaster or necessary outages for maintances.
Instructions for managing with Data Guard Broker, but also complete manual steps are given.
z
Learn how to setup a full automated failover solution, which will let SAP application reconnect to
database without doing anything manual, using Data Guard feature “Fast-Start failover” in
conjunction with “Client connection timeout” and database role event.
The topics contain cross-references to the appropriate chapter in the Annex.
The Annex is a collection of all the scripts and configuration commands you will have to write.
Texts in [] are optional.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Preparatory work
The work that must be carried out in advance is described below. This includes the database parameters,
SQLNet configuration and testing the SQLNet connection from the primary side to the standby host and vice
versa.
Naming conventions
Shown here taking the example of the O11 system:
As a prerequsite for using Data Guard Broker all service_names, global_db_names and
db_unique_name defined in listener.ora, tnsnames.ora and database parameter file has to be set up in
lower case.
Not following this prerequiste ends with configuration errors later on, as Data Guard broker convert
all input values like a mixed-case service_name (O11_oracle-lnx) to lower case (o11_oracle-lnx) . But
if the service_name is also defined in mixed_case at parameter file (O11_oracle-lnx), it will not be
found by Data Guard broker, as it acts case sensitive.
For example the domain is set to WORLD in the document, but it can be set by individual needs.
IDENTIFIER
PRIMARY DATABASE
PHYSICAL STANDBY DATABASE
COMMENTS
DB-SID
O11
O11
Listener.ora
Host name
oracle-lnx
oracle-vm2
Listener / tnsnames.ora
DG config.
o11
o11
Configuration name for
Broker
Service name
o11_oracle-lnx[.WORLD]
o11_oracle-vm2[.WORLD]
Tnsnames.ora
Listener name
Listener_o11
Listener_o11
Listener.ora
Listener port
1527
1527
Listener.ora
Global_Dbname
o11_oracle-lnx_DGMGRL [.WORLD]
o11_oracle-vm2_DGMGRL
Listener.ora
o11_oracle-lnx_DGB [.WORLD]
[.WORLD]
[11g] DGMGRL configured
o11_oracle-lnx_XPT [.WORLD]
o11_oracle-vm2_DGB [.WORLD]
in dg broker as property
(listener.ora) see
also SQL*Net
configuration
o11_oracle-vm2_XPT [.WORLD]
StaticConnectIdentifier
DGB and XPT will be autoconfigured o demand
DB_Unique_name
o11_north
o11_south
Spfile<sid>.ora
Important: define in lower
case!!
Oracle base
/oracle
/oracle
Oracle base
Data files
sapdata*
sapdata*
Data files
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Database parameters
Parameters in addition to those normally specified in SAP note "Parameter recommendations for Oracle 11g"
are needed to set up a Data Guard configuration between two databases. Below you will find a list of these
additional parameters for both the primary and the standby database.
Note: We would recommend configuring the Data Guard Broker (chapter Configuring Data Guard Broker),
then the parameters are set to the correct values depending on the properties configured once the Broker has
been started. This takes precedence over all values set manually up until that point for the parameters in
question.
If a Data Guard Broker is not being used, all parameters must be set manually in the parameter file. You will
find an example of an application in the next section
Definition of relevant parameters for Oracle Data Guard (see Oracle online documentation Data Guard
Concepts and Administration )
PARAMETER
DESCRIPTION
VALUE
log_archive_dest_1
Location where the archive logs are
Location=“<local directory>“
stored (locally)
[parameter] [valid_for =( expression )]
1. Destination for transferring logs to the
SERVICE=“<service definition>“
standby database. (remote)
[parameter] [valid_for =( expression )]
log_archive_dest_2
Other destinations are possible if
necessary and/or if needed for other
standby databases (1 – n)
log_archive_dest_state_n
Status of destinations, possible values:
ENABLE
ENABLE = active
DEFER = inactive
local_listener
Communicates the database's local
Name and definition in Tnsnames.ora
listener, needed for Data Guard
or as net service string in the form
communication
(ADRRESS= ( PROTOCOL= protocol)
(HOST= hostname) (PORT= port#) )
log_file_name_convert
Enables file name components to be
’String’ , ’String’
converted. In order to operate with
standby log files, this parameter must be
set to prevent unwanted log file clearing.
standby_file_management
Activates the automatic creation of new
data files which are produced on the
primary side during a tablespace
AUTO
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
PARAMETER
DESCRIPTION
VALUE
expansion.
OFF deactivates the feature
db_unique_name
Unique database name, important for
e.g. <dbsid>_<location>
identification because both databases are
usually addressed with the same SID.
Additional database parameters when operating with Data Guard Broker
PARAMETER
dg_broker_start
DESCRIPTION
VALUE
Starts or ends the DMON Data Guard
Manually:
Broker process
TRUE
TRUE = starts DMON
FALSE = stops DMON
When configuring the Data Guard Broker, the following parameters are automatically written to the
parameter file once the DMON process has been started:
log_archive_dest_2
1. Destination for transferring logs to the
SERVICE=“<service definition>“
standby database. (remote)
[parameter] [valid_for =( expression )]
Is generated in full from the Broker
configuration
log_archive_dest_state_n
Status of destinations, possible values:
ENABLE
ENABLE = active
DEFER = inactive
log_archive_config
Configuration name of Data Guard
Is created dynamically by the Data
configuration on the partner host.
Guard Broker
Other parameters are generated dynamically by the Data Guard Broker based un the settings of configuration
properties and placed in the SPFILE.
Defining the database unique name has an affect to the database diag directory. As of 11g the new diag feature
replaces the old background_dump_dest and user_dump_dest parameters. Alert.log and tracefiles will be
written in the diagnostic_dest subdirectories. Changing the database unique name in behalf of the Data Guard
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
configuration, a new subdirectory named with the db_uniquename will be created and all diagnostic output
will be redirected to this new folder.
Example of application for parameter settings:
Parameter Primary DB
*.db_unique_name='o11_north'
*.dg_broker_start=true
*.local_listener='(ADDRESS = (PROTOCOL = TCP)(HOST = oracle-lnx)(PORT = 1527))'
*.log_archive_dest_1='LOCATION="/oracle/O11/oraarch/O11arch",
valid_for=(ONLINE_LOGFILE,ALL_ROLES)'
*.log_archive_dest_state_1='ENABLE'
*.log_archive_format='%t_%s_%r.dbf'
*.log_archive_max_processes=2
*.log_archive_min_succeed_dest=1
*.log_archive_trace=0
*.log_file_name_convert='O11','O11'
*.standby_file_management='AUTO'
Parameter Standby DB
*.db_unique_name='o11_south'
*.dg_broker_start=TRUE
*.local_listener='(ADDRESS = (PROTOCOL = TCP)(HOST = oracle-vm2)(PORT = 1527))'
*.log_archive_dest_1='location="/oracle/O11/oraarch/O11arch" mandatory
valid_for=(all_logfiles,all_roles)'
*.log_archive_dest_state_1='ENABLE'
*.log_archive_format='%t_%s_%r.dbf'
*.log_archive_max_processes=2
*.log_archive_min_succeed_dest=1
*.log_archive_trace=0
*.log_file_name_convert='O11','O11'
*.standby_file_management='AUTO'
SQL*Net configuration
Data Guard needs a total of 3 different listener services per database.
SERVICE IDENTIFICATION
DESCRIPTION
SERVICE NAME USED
*_XPT
Used to transfer the redo logs and for FAL
P: o11_north_XPT[.WORLD]
requests
S: o11_south_XPT[.WORLD]
Is used by the background processes for
P: o11_north_DGB[.WORLD]
communication, e.g. for the heartbeat and
S: o11_south_DGB[.WORLD]
*_DGB
RFS communication.
*_DGMGRL
Used for switchover and failover by the
P: o11_north_DGMGRL[.WORLD]
Data Guard Broker. configured as property
S: o11_south_DGMGRL[.WORLD]
StaticConnectIdentifier
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
The *XPT and *DGB services are usually initiated automatically by the Data Guard Broker and created
dynamically, based on the database name given during create of configuration. This can be usually the same as
the db_unique_name setting at database.
The *_DGMGRL service must however be defined in the Listener.ora in all cases.
Data Guard Broker uses this service at switchover to connect to the stopped instances and startup it again. If
such a service would be dynamically (by instance) defined, it will not be available during instance shutdown.
Thus definition has to be static with the listener.
This service is defined as "Global_DBName" in the listener.ora.
Services XPT and DGB should be created automatically be Data Guard Broker, if not it should be created
manuall in LISTENER.ORA. See the example listener.ora in annex.
It may be platform dependend if service will be created by Data Guard Broker or not.
Possible errors
•
TNS-12514
”TNS: could not resolve the connect identifier specified"
*Cause:
A connection to a database or other service was requested using a connect identifier, and the connect
identifier specified could not be resolved into a connect descriptor using one of the naming methods
configured. For example, if the type of connect identifier used was a net service name then the net service
name could not be found in a naming method repository, or the repository could not be located or
reached.
*Action:
- If you are using local naming (TNSNAMES.ORA file):
- Make sure that "TNSNAMES" is listed as one of the values of the NAMES.DIRECTORY_PATH
parameter in the Oracle Net profile (SQLNET.ORA)
- Verify that a TNSNAMES.ORA file exists and is in the proper directory and is accessible.
- Check that the net service name used as the connect identifier exists in the TNSNAMES.ORA file.
- Make sure there are no syntax errors anywhere in the TNSNAMES.ORA file. Look for unmatched
parentheses or stray characters. Errors in a TNSNAMES.ORA file may make it unusable.
In order to prevent error ORA-12514 during a switchover, the DGMGRL service must be created manually
and more importantly without a domain even if the domain = WORLD is set. Should the ORA-12514 error
still arise, the service should be defined once with domain and once without domain in order to satisfy both
scenarios. For more details about the error, refer to the drc<sid>.log and sqlnet.log log files.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
•
TNS-12541
"TNS:no listener"
*Cause:
The connection request could not be completed because the listener is not running.
*Action:
Ensure that the supplied destination address matches one of the addresses used by the listener
- compare the TNSNAMES.ORA entry with the appropriate LISTENER.ORA file (or TNSNAV.ORA
if the connection is to go by way of an Interchange). Start the listener on the remote machine.
Check that the host name and port in the Listener.ora and tnsnames.ora configuration files match. Has the
listener been started?
The output:
Lsnrctl services <listener_name>
Provides an overview of the services initiated. The services produced by the Data Guard Broker are then also
listed.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Example application for an SQL*Net configuration:
Listener settings primary database
Listener Settings Standby DB
ADMIN_RESTRICTIONS_LISTENER_O11 = on
LISTENER_O11 =
(ADDRESS_LIST =
(ADDRESS =
(PROTOCOL = IPC)
(KEY = O11.WORLD)
)
(ADDRESS=
(PROTOCOL = IPC)
(KEY = O11)
)
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-lnx)
(PORT = 1527)
)
)
STARTUP_WAIT_TIME_LISTENER_O11 = 0
CONNECT_TIMEOUT_LISTENER_O11 = 10
TRACE_LEVEL_LISTENER_O11= OFF
SID_LIST_LISTENER_O11=
(SID_LIST =
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-lnx)
(ORACLE_HOME = /oracle/O11/11r2_32)
)
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_north_DGMGRL)
(ORACLE_HOME = /oracle/O11/11r2_32)
)
)
ADMIN_RESTRICTIONS_LISTENER_O11 = on
LISTENER_O11 =
(ADDRESS_LIST =
(ADDRESS =
(PROTOCOL = IPC)
(KEY = O11.WORLD)
)
(ADDRESS=
(PROTOCOL = IPC)
(KEY = O11)
)
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-vm2)
(PORT = 1527)
)
)
STARTUP_WAIT_TIME_LISTENER_O11 = 0
CONNECT_TIMEOUT_LISTENER_O11 = 10
TRACE_LEVEL_LISTENER_O11= OFF
SID_LIST_LISTENER_O11=
(SID_LIST =
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-vm2)
(ORACLE_HOME = /oracle/O11/11r2_32)
)
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_south_DGMGRL)
(ORACLE_HOME = /oracle/O11/11r2_32)
)
)
Tnsnames.ora primary database
Tnsnames.ora Standby DB
O11_oracle-lnx.WORLD=
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-lnx)
(PORT = 1527)
)
)
(CONNECT_DATA =
(SID = O11)
(GLOBAL_NAME = O11)
(SERVICE_NAME = o11_oracle-lnx)
)
)
O11_oracle-vm2.WORLD=
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-vm2)
O11_oracle-vm2.WORLD=
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-vm2)
(PORT = 1527)
)
)
(CONNECT_DATA =
(SID = O11)
(GLOBAL_NAME = O11)
(SERVICE_NAME = o11_oracle-vm2)
)
)
O11_oracle-lnx.WORLD=
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-lnx)
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
(PORT = 1527)
(PORT = 1527)
)
)
)
(CONNECT_DATA =
(SID = O11)
(SERVICE_NAME = o11_oracle-lnx)
)
)
(CONNECT_DATA =
(SID = O11)
(SERVICE_NAME= o11_oracle-vm2)
)
)
)
Checking the SQL*Net configuration
Once the configuration files have been created and implementation on the hosts is complete, you must check
whether the databases can be accessed by the other partner host using SQL*Net. The TNSPING Oracle tool
is used for this purpose.
tnsping <DB-Name>
lsnrctl services <listener_name>
Example:
DB-Name here corresponds to the identifier created in the tnsnames.ora for the database.
PRIMARY HOST
STANDBY HOST
tnsping o11_oracle-vm2[.WORLD]
tnsping o11_oracle-lnx[.WORLD]
lsnrctl services Listener_O11
lsnrctl services Listener_O11
A valid SQL*Net address string must be returned and must end with details of the runtime. If errors such as
TNS-12541 "No listener" occur, this indicates incorrect configuration.
Then first check
•
Whether the listener on the standby host has been started,
•
Whether the host name is written correctly and can also be canceled
•
Whether the port in Tnsnames.ora and the Listener.ora are the same.
•
The service name used is identical.
Corrections can be undertaken on the basis of the error message.
Structure of the standby database
The database is usually produced from an OFFLINE or even an ONLINE backup of the production
database. There are many ways of doing this but it can be easily done using e.g. the RMAN.
You will find the official Oracle description of how to set up a Data Guard environment in the Oracle
documentation: Oracle Database High Availability Best Practices 10g Release 2 - Documentation0.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Copying the production database (sample scenario)
In this scenario the datafile will be copied using ONLINE BACKUP feature with filesystem copies.
Its also possible to do this if database is down, in that case “Begin Backup” and “End Backup” commands
can be skipped.
Carry out on: oracle-lnx
SQL> alter database begin backup;
Copy data. This is done e.g. with the database user and SCP.
scp -r /oracle/O11/sapdata* oracle@oracle-vm2:/oracle/O11/*
Carry out on: oracle-vm2oracle-lnx
SQL> alter database end backup;
As a rule when creating database files on the standby system you can use the same method as you would for a
homogeneous system copy. Other possibilities are to:
•
Restore a backup from tape
•
Copy the database using RMAN
Copy database parameter file and password file
•
Copy the database parameter file init<sid>.ora to the standby Oracle Home destination:
$ORACLE_HOME/dbs
Change all the parameters like given in the example found in annex.
•
Copy the passwordfile : orapw<SID> to standby Oracle Home destination:
$ORACLE_HOME/dbs
The password can not be changed at standby side, as the database will be normally not opened.
Force logging
The most SAP databases are running with activated redo logging mode, as recommended, so all changes will
be logged and transported to the Standby DB by logfiles.
For SAP BI systems there is an exception regarding the secondary indexes for the info cubes. All these
indexes are created using the nologging option, therefore for these segments no redo will be written to
logfiles. As an effect at the recovery at Standby DB, these blocks are marked as soft corrupt and will be
skipped.
If standby database will become primary this will throw corruption errors during open.
An easy method to workaround this is by setting primary in permanent FORCE LOGGING mode, but it will
have a negative effect for performance during the load phases of info cubes.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
At the other hand, recreating all the secondary indexes in case of the role switch is needed, but will need time
before the database will become productive.
If a standby database is used for a SAP BI system, its on the customer to decide what way would be the best
in his situation.
For OLTP organized systems FORCE LOGGING is also an option, but not really needed, if there will be
make no use from the nologging option.
To set a primary database in FORCE LOGGING mode use:
ALTER DATABASE FORCE LOGGING;
Creating a standby control file on the primary database
Carry out on: oracle-lnx
SQL> startup mount
SQL> alter database create standby controlfile as '/oracle/O11/sapreorg/stdby_controlf.dbf' reuse;
You will find a suitable script in the Annex (cre_stdbycontrolfile.sql).
But the command can also be run directly using SQLPLUS. The standby control file created now has to be
copied to the standby host.
Copy data. This is done e.g. with the database user and SCP.
scp -r /oracle/O11/sapreorg/stdby_controlf.dbf
vm2:/oracle/O11/sapreorg
oracle@oracle-
Distributing on the standby host
For example:
cp /oracle/O11/sapreorg/stdby_controlf.dbf /oracle/O11/origlogA/cntrl/cntrlO11.dbf
cp /oracle/O11/sapreorg/stdby_controlf.dbf …
cp /oracle/O11/sapreorg/stdby_controlf.dbf …
The file is copied into the directories which are stated in the database parameter file using the
CONTROLFILES= parameter. The example in the Annex shows one possible process using shell script
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Mounting the standby database for the first time
This is done to test the function. It tests whether all database parameters match and whether the control files
are in the right place and are readable. Also we can now take the chance to create a SPFILE from the copied
init<SID>.ora.
SQL> connect / as sysdba
SQL> create spfile from pfile;
SQL> startup mount
The Data Guard Broker starts up but doesn't find a valid configuration, as a result of which error messages
are logged in Alert.log and in Broker.log (drc<sid>.log).
For the remaining configuration period, it is very helpful to execute a tail -f on the alert log of both databases.
This allows any errors to be detected and remedied immediately.
Creating standby redo log files
If the Data Guard logs are to be transferred with LGWR in ASYNC or SYNC mode, standby redo log files
are needed. These receive the redo information "in time" from the primary database and buffer it. This
ensures that if the online redo logs are lost on the primary side, all redo information is already present on the
standby side.
If ARCH is used instead, standby redo log files do not have to be created, and if they have already been
produced, they are not used.
The number of standby redo log files must always be 1 higher than the number of log groups defined. Should
the amount of redo operations increase, this prevents the redo information from not being updated because
there are no more free standby log files available. The size of the standby redo log files must match that of the
normal redo log files. Therefore, please adjust the script in the Annex accordingly before using it to create the
standby log files.
Standby log files should be produced on the standby side and on the primary side. The latter are needed in the
event of a switchover.
Make sure that the directory exists on the file system.
Start SQLPLUS "connect / as sysdba"
Syntax:
SQL> ALTER DATABASE ADD STANDBY LOGFILE '<FILE DESTINATION>' SIZE <n>M REUSE;
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Standby log files
Example:
[11g]
mkdir
SQL>
alter
alter
alter
alter
alter
replace
/oracle/O11/standbylog
database
database
database
database
database
add
add
add
add
add
standby
standby
standby
standby
standby
logfile
logfile
logfile
logfile
logfile
'/oracle/O11/standbylog/srl1.dbf'
'/oracle/O11/standbylog/srl2.dbf'
'/oracle/O11/standbylog/srl3.dbf'
'/oracle/O11/standbylog/srl4.dbf'
'/oracle/O11/standbylog/srl5.dbf'
size
size
size
size
size
100M
100M
100M
100M
100M
reuse;
reuse;
reuse;
reuse;
reuse;
Configuring Data Guard Broker
Open primary database and mount standby database. The DMON process (Data Guard Monitor) on both
database get active (dg_broker_start=TRUE) .
Make sure that the listener on the standby database is started and that the database can be accessed with
SQL*Net. (See Checking the SQL*Net configuration)
Any Error messages relating to the LOG_ARCHIVE_DEST_2 parameters can be prevented with
SQL> ALTER SYSTEM SET log_archive_dest_state_2='DEFER' SCOPE=MEMORY;
if log_archive_dest_2 has already been manually entered in the parameter file of the database.
To prevent any conflict with the automated broker settings, all manually set parameter for running a standby
database should be removed from spfile before setup a broker configuration.
Using the Data Guard Broker the log_archive_dest_2 parameter is written to the SPFILE triggered by the
Broker configuration.
Defining the protection mode
The transfer mode for the redo log files depends on the protection mode selected. Also refer to the
documentation "Managing Data Protection mode".
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
First of all you should find out which protection mode Data Guard should be used in. Unless configured
otherwise, the "Maximum Performance" default is used.
For the setup and testing phase “Maximal performance” mode is the best option. Later on, if the
configuration works it can be changed for the configuration by particulary broker command. See Data Guard
Broker Reference documentation for details.
The following modes are supported with SAP.
•
Maximum Performance
Maximum Performance mode is the default protection mode. It offers slightly less primary database data
protection, but higher performance than Maximum Availability mode. In this mode, as the primary
database processes transactions, redo data is asynchronously shipped to the standby database by the
LGWR process. Alternatively, the Archiver process(es) (ARCH) on the primary database may also be
configured to transmit the redo data in this mode. In any case, the commit operation of the primary
database does not wait for the standby database to acknowledge receipt before completing the write
process on the primary database. If any standby destination becomes unavailable, processing continues on
the primary database and there is little or no effect on performance.
In the case of failure of the primary database, redo data, which had not yet been sent to the standby
database, is lost. However, if the network has sufficient throughput to keep up with peaks in redo traffic
and the LGWR process is used to transmit redo to the standby server, the number of lost transactions
will be very small or zero.
The Maximum Performance mode should be used when availability and performance on the primary
database are more important than the risk of losing a small amount of data. This mode is also suitable for
Data Guard deployments over a WAN where the inherent latencies of the network may limit the
suitability of synchronous redo transmission.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
•
Maximum Availability
Maximum Availability mode offers the next highest level of data availability for the primary database. As
with Maximum Protection mode, redo data is synchronously transmitted by LGWR from the primary
database to the standby database, and the transaction is not committed on the primary database until it
has been confirmed that the transaction data is available on disk on at least one standby server. However,
in this mode, unlike the Maximum Protection mode, if the last participating standby database becomes
unavailable e.g. because of network connectivity problems, processing continues on the primary database.
The standby database may temporarily fall behind compared to the primary database, but when it is
available again, the databases will automatically synchronize with no data loss, using accumulated archived
logs on the primary database. Because of synchronous redo transmission, this protection mode can
potentially impact on response time and throughput. Configuring a low latency network with sufficient
bandwidth for peak transaction loads can minimize this impact.
Maximum Availability mode is suitable for businesses that want the assurance of zero data loss
protection, but do not want the production database to be impacted on by network/standby server
failures. These businesses will accept the possibility of data loss should a second failure subsequently
affect the production database before the initial network/standby failure is resolved.
•
Maximum Protection
Maximum Protection mode offers the highest level of data protection for the primary database, ensuring
a comprehensive zero-data loss disaster recovery solution. When operating in Maximum Protection
mode, redo records are synchronously transmitted by the LGWR process (through the LNS process)
from the primary database to the standby database(s), and a transaction is not committed on the primary
database until it has been confirmed that the transaction data is available on disk on at least one standby
server. It is strongly recommended that this mode be configured with at least two standby databases. If
the last participating standby database becomes unavailable, processing stops on the primary database.
This ensures that no transactions are lost should the primary database fail after it loses contact with all of
its standby databases.
Because of the synchronous nature of redo transmission, this Maximum Protection mode can potentially
impact on primary database response time. Configuring a low latency network with sufficient bandwidth
for peak transaction load can minimize this impact. Stock exchanges, currency exchanges, and financial
institutions are examples of businesses that may require this Maximum Protection mode.
"Maximum Performance" mode is recommended for operating a standby database with SAP when using
LGWR. The Data Guard Broker configuration described here is therefore based on this recommendation.
"Maximum Availability" mode is also available. This mode offers increased data security but decreases
performance. The mode automatically switches back to "Maximum Performance" mode if synchronous
communication cannot be maintained. “Max Availability Mode” is required if you want to setup Fast-Start
Failover feature.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Delayed Redo apply or Real time apply
In the most of all cases at SAP, a delayed redo apply would be the best decision, as it will provide the best
protection for human errors. So a delay of 3-4 hours protects the Standby DB for applying one or more
destructive command to fast and give DBA a good chance to react by stopping the log apply and then extract
unchanged data from the Standby database. This is one of the big advantages using a Standby database.
The amount of minutes to delay in past is defined by Data Guard Broker property DelayMins, for details see
next chapter. How large the delay can be depends on
•
the amount of redo created in a time window.
•
the amount of redo can be applied in a time window.
•
and the maximal time allowed for switch or failover which is defined by business needs.
Alternatively a Standby database can run in real time apply mode, that means, that all changes from the
Primary database are immediately applied to Standby database, if the archive log would be recovered, At real
time apply, this happens at least having the next logswitch at Primary.
Real time apply makes only sense in conjunction with configured and activated flashback database feature at
the Standby database. Only then the database can be set back to the past again to retrieve data in case of
human errors.
Real time apply is required for using Fast-Start failover.
For setup a Standby database with flashback database feature see Oracle White Paper :
”Setup Flashback Database on Data Guard Physical Standby Database for SAP Customers”
which you can find at http://www.oracle.com/sap under ORACLE PRODUCTS FOR SAP Æ Oracle
Database Æ Best Practice
Creating the configuration
The DGMGRL Broker interface should be started as the Oracle user:
oracle-lnx> DGMGRL
DGMGRL> connect sys
DGMGRL> password:
DGMGRL> create configuration <DG-Config-Name> as primary database is
<DB_Unique_Name> connect identifier is <Service_name>;
Creating the Data Guard configuration
Example:
DGMGRL> create configuration o11 as primary database is o11_primary connect identifier is
o11_oracle-lnx;
Configuration "o11" created with primary database "o11_primary"
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Adding the physical standby database
The standby database must be mounted for this purpose.
DGMGRL> add database <DB_unique_Name> as connect identifier is <Service_name>
maintained as physical;
Add physical standby database to DG config.
Example:
DGMGRL> add database o11_standby as connect identifier is o11_oracle-vm2 maintained as
physical;
Database "o11_standby" added
Adapting the Data Guard Broker properties
The Data Guard Broker needs several settings relating to the databases, which deviate from the preset default
settings. It is important that these properties match the corresponding database parameters, in case such
parameters are already set.
A good way to make the configuration first time runnable is to use the defaults at first hand, and change the
parameter needed later on using broker commands.
PROPERTY NAME
CORRESP. DATABASE PARAMETER
COMMENTS
LogXptMode
- within log_archive_dest_n
ASYNC should always be used when the LGWR is
Definition.
used to transfer logs in "Maximum Performance"
mode.
(This is the default)
SYNC is needed for operation with "Maximum
Protection" or "Maximum Availability". Drops in
performance on the primary databases are possible.
ARCH is used for operation with "Maximum
Performance" mode and transfer of logs using the
ARCH process.
DelayMins
MaxFailure
logFileNameConvert
- within log_archive_dest_n
Length of delay on standby side in minutes
Definition.
Defaullt = 0
- within log_archive_dest_n
Standard error tolerance for connection errors before
Definition.
an abort takes place. Default = 0
Log_file_name_convert
Default = '<SID>','<SID>'
will be imported from database.
hostname
-
use the physicaly hostname
LocalListenerAddress
Local_listener
removed with 11.2
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Data Guard properties
Example:
Set the Delay for applying the redo to 3 hours:
DGMGRL> EDIT DATABASE o11_primary SET PROPERTY DelayMins='180';
DGMGRL> EDIT DATABASE o11_standby SET PROPERTY DelayMins='180';
Activating the configuration
The configuration must be activated before the Data Guard Broker can use it.
DGMGRL> ENABLE CONFIGURATION;
Errors may arise as a result of which the configuration cannot be activated. The Broker log file provides a
check for this under:
/oracle/<SID>/saptrace/background/drc<sid>.log
Upon successful activation, the Data Guard Broker now undertakes the final parameterization for the
databases on the basis of the properties and activates media recovery on the standby database in the
background.
Æ From this point on Data Guard is active.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Deploying the transfer of logs and function check
When using a Data Guard Broker configuration, the transfer of logs and the media recovery run without any
user intervention.
To start the Data Guard and get it running, all that is required is:
•
To set dg_broker_start=TRUE on both databases
•
For the primary database to be open and
•
For a standby database to have been mounted
The properties in the Broker can be changed at any time via the DGMGRL Broker interface. Please read the
Oracle documentation Data Guard Broker Administration for more details.
Carry out a few manual log switches to perform the function check:
Alter system switch logfile;
Also check whether the log switch causes the redo to be transferred to the standby database. The Alert.log for
both databases filled with corresponding entries.
On the primary database:
Wed Jan 20 22:09:07 2010
Thread 1 advanced to log sequence 151 (LGWR switch)
Current log# 7 seq# 151 mem# 0: /oracle/O11/origlogA/log_g14m1.dbf
Wed Jan 20 22:09:43 2010
LNS: Standby redo logfile selected for thread 1 sequence 151 for
destination LOG_ARCHIVE_DEST_2
On the standby database:
Wed Jan 20 22:14:45 2010
Media Recovery Waiting for thread 1 sequence 151 (in transit)
Thu Jan 21 05:37:52 2010
Archived Log entry 190 added for thread 1 sequence 151 ID 0x681c76d7 dest
3:
ARC0: Archive log thread 1 sequence 151 available in 5 minute(s)
Thu Jan 21 05:37:55 2010
Media Recovery Delayed for 5 minute(s) (thread 1 sequence 151)
Once the delay has passed, the media recovery should also be visible:
Thu Jan 21 05:42:56 2010
Media Recovery Log <logfile name>
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Implementing SAP BRARCHIVE with Data Guard
As usual at SAP systems SAP tool BRARCHIVE is used for storing the archive logs to tape and clean up the
archive log destination to prevent an archiver stuck. All archive logs which are written to log_archive_dest_1
will be handled by BRARCHIVE that way.
There is no change to BRARCHIVE configuration necessary. It can work like with any other standolane
database.
Remember the way how redo information are processed since Oracle 10g Data Guard. As described, we now
are using LGWR , what give some advantages compared using the ARCH process.
All redo will be immediately written to Standby Database log files using log_archive_dest_2 and therefore all
that redo will be already available at Standby host. Further handling now is completely at the Standby
Database, having a log switch at Primary will also switch the Standby logfile and then archive the former
standby logfile to an archive file at Standby Database.
Such files will be applied by recover process (MRP) if the delay time is reached.
But what happens in the case Standby Database missed some standby logs and therefore even the standby
archive log due to maintenance or temporary offline state of Standby Database?
The normal reaction of Standby Database is to request missing archive files, based on the log sequences from
the Primary Database again. Unfortunately the files are already stored to tape and deleted by BRARCHIVE,
so the ongoing FAL-Request will fail.
The missed files have to be restored back from tape again to log_archive_dest_1. This is time consuming and
has to be done manually be DBA.
Using an additional Buffer destination
In case of a FAL-Request, Primary Database will check all its locally destinations for the requested files and if
successfully send it again or returns an error if not found.
The idea is now, to define a secondary locally destination in log_archive_dest_3 to buffer all created archive
files for the time of up to 24 hours.
That destination is not touched by BRARCHIVE. So the file should be still exists here, and FAL-Request can
find and send the file from this secondary destination and no further manual action is needed.
Define this destination at Primary database like:
*.log_archive_dest_3='LOCATION="/oracle/PEC/oraarch2/PECarch", optional,
valid_for=(ONLINE_LOGFILE,ALL_ROLES)'
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
For preventing any archiver stuck due to file system full at secondary destination, the parameter OPTIONAL
has to be set.
Starting the time of definition of the additionally locally destination primary ARCH process will write an
archivelog file copy in parallel to this buffer destination too.
Maintance the buffer destination
The destination should be able to hold archive log files for the last 24 hours. How large the filesystem has to
be depends on the size and amount of archive logs created for that timeframe.
In some cases 24 hours would be too long and it will be needed too much disk space to hold all the files,
therefore decide what is the best solution for your needs and money.
Calculate it like:
Destination size = (number of archives * size) / hour * 24
Create a independent filesystem which the given size. It can optional reside at any NFS share if necessary.
Implement a rolling delete routine, may be a cron job, for delete all files which are older then the given time
(e.g. 24h) and already applied at Standby Database.
Using:
SQL> SELECT THREAD#, MAX(SEQUENCE#) AS "LAST_APPLIED_LOG"
FROM V$LOG_HISORY GROUP BY THREAD#;
To find the last applied log from Standby Database to take care that the log can be deleted from destination.
Taking backups from Standby Database using BRBACKUP
From the Standby Database we can perform OFFLINE backups without having any impact to performance
at the primary side. You can use BRBACKUP in a similar way as using it for the primary side. BRBACKUP
only needs to know that is running at a Standby Database, so its logging and controlling tables are still written
to primary side.
In configuration file init<sid>.sap at $ORACLE_HOME/dbs add the parameter
Backup_type = offline_standby
For more information see:
http://help.sap.com/saphelp_40b/helpdata/de/0d/d311854a0c11d182b80000e829fbfe/content.htm
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Disaster handling / role transition
Preparations
Also refer to "Data Guard Concepts and Administration " chapter 8.1.1 Preparing for a Role Transition
(failover or switchover).
Verify that the initialization parameters for each database are configured correctly.
•
The database, which takes on the role of the primary database, must run in ARCHIVELOG mode.
•
Make sure that all files, which belong to a tablespace with temporary files, also exist on the standby side
before the role transition takes place.
How long does the role transition take?
Data Guard provides the V$DATAGUARD_STATS view, which can estimate how long the process, will
take.
Example:
SQL> COLUMN NAME FORMAT A18
SQL> COLUMN VALUE FORMAT A16
SQL> COLUMN TIME_COMPUTED FORMAT A24
SQL> SELECT * FROM V$DATAGUARD_STATS;
NAME
VALUE
UNIT
TIME_COMPUTED
---------------------- ---------------- ------------------------------ ----------------------apply finish time
apply lag
+00 00:15:51
estimated startup time 10
standby has been open
transport lag
day(2) to second(1) interval
19-MAY-2008 09:29:07
day(2) to second(0) interval
19-MAY-2008 09:29:07
second
19-MAY-2008 09:29:07
N
+00 00:00:00
19-MAY-2008 09:29:07
day(2) to second(0) interval 19-MAY-2008 09:29:07
The example shows that there is no delay in the transfer of redo logs, but this is of importance only if another
standby database is being cascaded. The active media recovery currently needs another 15:51 minutes.
The role transition will only take place once it is complete.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Switchover
During a switchover the roles of the databases are usually swapped. The previous standby database becomes
the primary database and the previous primary one the standby one.
This is done by parameterization and changing the control file.
Data Guard before a switchover
After a switchover
During the switchover phase both databases briefly assume the standby role.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Switchover with Data Guard Broker
The switchover is triggered by a simple command in the DGMGRL tool.
(“Data Guard Broker” chapter 6.8 scenario 7: Performing a Switchover Operation.)
It is important that the connect is made using a valid Listener Service which is statically defined in listener.ora
and is defined in the local tnsnames.ora . For example “o11_oracle-lnx” service.
In case of a switchover connection can be made to the local DB service, here “o11_oracle-vm2” unique
database name O11_SOUTH, or alternatively to the remote service “o11_oracle-lnx” unique database name
O11_NORTH
DGMGRL> CONNECT SYS@o11_oracle-vm2
DGMGRL> Password: ***********
DGMGRL> show configuration
Configuration - o11
Protection Mode: MaxPerformance
Databases:
o11_south - Primary database
o11_north - Physical standby database
Fast-Start Failover: DISABLED
Configuration Status:
SUCCESS
DGMGRL> SWITCHOVER TO o11_north;
Switchover without Data Guard Broker
This approach is only needed if the Data Guard Broker has not been used. The steps have to be processed
manually following a list. Very experienced database administrators should only undertake this approach, as it
is very prone to error and can therefore results in problems when transitioning roles. This may even extend to
scenarios where it is no longer possible to synchronize the standby database.
Manual approach
Step 1
Verify that it is possible to perform a switchover.
On the current primary database, query the SWITCHOVER_STATUS column of the
V$DATABASE fixed view on the primary database to verify it is possible to perform a switchover. For example:
SQL> SELECT SWITCHOVER_STATUS FROM V$DATABASE;
SWITCHOVER_STATUS
-----------------
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
TO STANDBY
1 row selected
Step 2
Initiate the switchover on the primary database.
SQL> ALTER DATABASE COMMIT TO SWITCHOVER TO PHYSICAL STANDBY
[WITH SESSION SHUTDOWN];
Step 3
Shut down and restart the former primary instance.
Shut down the former primary instance, restart and mount the database:
SQL> SHUTDOWN IMMEDIATE;
SQL> STARTUP MOUNT;
At this point in the switchover process, both databases are configured as physical standby
databases.
Step 4
Verify the switchover status in the V$DATABASE view.
SQL> SELECT SWITCHOVER_STATUS FROM V$DATABASE;
SWITCHOVER_STATUS
----------------TO_PRIMARY
1 row selected
Any other value for the switchover_status will prevent a switchover. Follow the troubleshooting guide in “Oracle Data Guard
Concepts and Administration” in this case.
Step 5
Switch the target physical standby database role to the primary role.
SQL> ALTER DATABASE RECOVER MANAGED STANDBY DATABASE CANCEL;
SQL> ALTER DATABASE COMMIT TO SWITCHOVER TO PRIMARY;
Step 6
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Finish the transition of the standby database to the primary role.
The task you perform depends on whether the physical standby database has ever been opened in read-only mode:
•
If the physical standby database has not been opened in read-only mode since the last time it was started, issue the SQL
ALTER DATABASE OPEN statement to open the new primary database:
SQL> ALTER DATABASE OPEN;
Then, go to step 7.
•
If the physical standby database has been opened in read-only mode since the last time it was started, you must shut down the
target standby database and restart it:
SQL> SHUTDOWN IMMEDIATE;
SQL> STARTUP;
Step 7
If necessary, restart log apply services on the standby databases. For the new physical standby database and for each other physical
or logical standby database in the Data Guard configuration, if log apply services were not previously configured to continue
operating through a switchover, use an appropriate command to restart log apply services.
Step 8
Begin sending redo data to the standby databases. Issue the following statement on the new primary database:
SQL> ALTER SYSTEM SWITCH LOGFILE;
Failover to standby database
Limitation:
This step must be taken when the primary system can no longer be accessed.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Primary database failure
All redo transactions, which have not been written to the standby log files or are available, as an archive log
cannot be recovered.
•
If the database was operated in "Maximum Protection" or "Maximum Availability" mode, it is very
likely that this will only affect the very last transaction which was not committed on the primary system
and can therefore be considered as not having been written.
•
When in "Maximum Performance" mode, this affects all transactions, which have not been transferred
to the standby, redo logs since the last net buffer synchronization using the RFS processes running on the
standby system.
•
If the logs were transferred using ARCH, the last complete archive log and the online redo logs on the
standby side will also be missing after the disaster.
•
If log transfer was configured with LGWR, usually all REDO information produced before the disaster
has been written to the standby redo log files.
•
Check whether the online redo logs or archive logs on the primary database server are still readable and
can be manually copied to the standby server.
Failover with Data Guard Broker
When using the DGMGRL tool, a failover can be carried out with just one command. Also refer to (“Data
Guard Broker” chapter 6.9 Scenario 8: Performing a Manual Failover Operation)
DGMGRL> CONNECT SYS@o11_oracle-lnx
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
DGMGRL> Password ********
DGMGRL> SHOW CONFIGURATION
DGMGRL> FAILOVER TO o11_north;
Failover without Data Guard Broker
This approach is only needed if the Data Guard Broker has not been used. The steps have to be processed
manually following a list. Very experienced database administrators should only undertake this approach, as it
is very prone to error and can therefore results in problems when transitioning roles. This may even extend to
scenarios where it is no longer possible to synchronize the standby database.
Manual approach
Step 1
Identify and resolve any gaps in the archived redo log files. To determine if there are gaps in the archived redo log files on the target
standby database, query the V$ARCHIVE_GAP view.
For example:
SQL> SELECT THREAD#, LOW_SEQUENCE#, HIGH_SEQUENCE# FROM V$ARCHIVE_GAP;
THREAD#
LOW_SEQUENCE# HIGH_SEQUENCE#
---------- ------------- -------------1
90
92
Copy all missing log files (90,91,92) to log destination manually.
Register them:
SQL> ALTER DATABASE REGISTER PHYSICAL LOGFILE 'filespec1';
Step 2
Repeat Step 1 until all gaps are resolved.
Step 3
Copy any other missing archived redo log files.
SQL> SELECT UNIQUE THREAD# AS THREAD, MAX(SEQUENCE#)
OVER (PARTITION BY thread#) AS LAST from V$ARCHIVED_LOG;
THREAD
LAST
---------- ---------1
Step 4
100
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Initiate a failover on the target physical standby database. Issue the following statement to initiate the failover:
SQL> ALTER DATABASE RECOVER MANAGED STANDBY DATABASE FINISH FORCE;
J FORCE keyword terminates active RFS processes on the target physical standby database
Note:
Failover adds an end-of-redo marker to the header of the last log file being archived and sends
the redo to all enabled destinations that are valid for the primary role (specified with the
VALID_FOR=(PRIMARY_ROLE, *_LOGFILES) or the VALID_FOR=(ALL_ROLES, *_LOGFILES)
attributes).
Note:
The FINISH keyword must follow all other keywords in the SQL statement, except for FORCE, WAIT, or NOWAIT.
Step 5
Convert the physical standby database to the primary role.
After the recovery has finished, change the role of database to primary:
SQL> ALTER DATABASE COMMIT TO SWITCHOVER TO PRIMARY;
Step 6
Finish the transition of the standby database to the primary database role. The task you perform in this step depends on whether
the physical standby database was ever opened in read-only mode:
•
If the physical standby database has not been opened in read-only mode since the last time it was
started, issue the SQL ALTER DATABASE OPEN statement to open the new primary database:
SQL> ALTER DATABASE OPEN;
Then, go to step 7.
•
If the physical standby database has been opened in read-only mode since the last time it was started, you must shut down the
target standby database and restart it:
SQL> SHUTDOWN IMMEDIATE;
SQL> STARTUP;
The target physical standby database has now undergone a transition to the primary database role.
Step 7
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Back up the new primary database. If you choose online backup, set database in backup mode:
ALTER DATABASE BEGIN BACKUP;
Once the backup has finished, don't forget to end backup mode with:
ALTER DATABASE END BACKUP;
Restoring the standby database after a failover
If the primary database host can be used again, there are various ways of moving the database back onto the
computer. You can use an offline backup from tape or a normal online backup via the file system. Oracle 10g
also provides features for an automatic "Reinstate" command in the DGMGRL. However, this option is not
available when using physical standby databases.
The procedure is the same as that used when creating a standby database.
•
Set up using an online backup of the current production database
•
Create a standby control file, copy and distribute to the new production host
•
Renew the database property in the Data Guard Broker configuration:
DGMGRL> connect sys/<password>
DGMGRL> show configuration
DGMGRL> remove database <database name>
DGMGRL> add database <DB_unique_Name> as connect identifier is <Service_name>
maintained as physical;
DGMGRL> EDIT DATABASE <DB_unique_Name> set PROPERTY ...
Also see:
•
Add physical standby database to Data Guard configuration.
•
Data Guard properties
•
The database service must also be recreated for Windows systems.
If the database has been recreated in this way, it is started up as a "physical standby database" and
synchronized with the current production database with a normal delay until a scheduled downtime permits
another switchover to return the database to its old role.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Fast-Start Failover
Definition
Fast-Start Failover enables an automated failover to standby database, in case the primary database goes down
by incident or network loss. An Observer process is used to monitor the network connectivity and availability
of the databases. The observer is a separate OCI client-side component that runs on a different computer
from the primary and standby databases and monitors the availability of the primary database.
In case the connection to primary database got lost and does not come back in a defined timeout range,
observer initiate a failover to standby database.
In difference to a manual failover, primary database will be reinstate by observer process automatically as a
new standby database. No manual work is necessary by the DBA.
Later on, a gracefull planed switchover during a scheduled downtime will bring back the databases into their
old roles.
See Oracle documenatation Data Guard Broker - Fast-Start Failover chapter 5.5 for details.
•
before Fast-Start failover
Observer holds and monitor connections to primary and standby database. Primary is sending its
redo to standby database.
•
Fast-Start Failover enuses
Observer has lost the connection to primary, even standby database too, so Fast-Start Failover is
triggered. Standby database becomes the new primary.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
•
After Fast-Start Failover
Restoring the old primary database and bring it back into Mount state, enables observer to connect
successfully again. Old primary will now reinstate as a standby database, triggered by observer
process.
Events, which will trigger a Fast-Start failover and which are enabled by default are:
•
Broken network connection between the observer and the primary database.
If the connection is lost between the observer and the primary database, or there are network failures
that cause the primary database to be isolated, the observer attempts a fast-start failover.
•
Instance failures
If a single-instance primary database (either RAC or nonRAC), or if all instances of a RAC primary
database fail, the observer attempts a fast-start failover.
•
Shutdown abort
If a single-instance primary database (either RAC or nonRAC), or if all instances of a RAC primary
database are shut down with the ABORT option, the observer attempts a fast-start failover. Fast-start
failover will not be attempted for the other types of database shutdown (NORMAL, IMMEDIATE,
TRANSACTIONAL).
•
Offline datafiles
If the observer determines that one or more datafiles in the primary database have been taken offline
by the database because of I/O errors, the observer attempts a fast-start failover.
•
Corrupted Dictionary
Dictionary corruption of a critical database object. Currently, this state can be detected only when the
database is open
•
Corrupted Controlfile
Controlfile is permanently damaged because of a disk failure.
Optional following events can be enabled too:
•
Inaccessible Logfile
LGWR is unable to write to any member of the log group because on an I/O error
•
Stuck archiver
Archiver is unable to archive a redo log because the device is full or unavailable.
Except for the last condition (Offline datafiles), the observer attempts to reconnect to the primary database
within the time specified by the FastStartFailoverThreshold configuration property before attempting a faststart failover. When the primary database datafiles are offline, the observer initiates a fast-start failover
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
immediately, without waiting for the amount of time specified by the FastStartFailoverThreshold property to
expire.
Prerequistites
The following prerequisites must be met before the broker allows you to enable fast-start failover
•
Ensure the broker configuration is running in either maximum availability mode or maximum
performance mode.
(See “Data Guard Broker” Set the protection mode for configuration Section 4.6.1 for information
about configuring the protection mode, standby redo logs, and the LogXptMode property.)
•
running real time apply is no longer a prerequisite with 11g, but recommended to prevent warnings
about fast_start failover is lagging.
Enable Flashback Database and set up a flash recovery area on both the primary database and the target
standby database. This allows to jump back in time if necessary, so a delayed redo apply at standby database is
not necessary and we can use real time apply .
See Oracle Whitepaper Setup Flashback Database on Data Guard Physical Standby Database for SAP
Customers or Oracle documentation "Setting Up Flash Recovery Areas as Destinations" in Data Guard
Concepts and Administration
•
Install the DGMGRL command-line interface on the observer computer as described in “Data
Guard Broker” - Oracle Data Guard Installation, Section 2.1.
•
Configure the TNSNAMES.ORA file on the observer system so that the observer is able to connect
to the primary database and the pre-selected target standby database.
Setup
See Oracle documentation “Data Guard Broker”- Enabling Fast-Start Failover, section 5.5.2.
Step 1: Configure and activate Flashback database
Follow the steps provided in Setup Flashback Database on Data Guard Physical Standby Database for SAP
Customers. A Flashback area need to be set up at Primary and Standby database.
Step 2: Set the protection mode
Find out what protection mode is currently used:
SELECT PROTECTION_MODE, PROTECTION_LEVEL FROM V$DATABASE;
PROTECTION_MODE
PROTECTION_LEVEL
-------------------- --------------------
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
MAXIMUM PERFORMANCE
MAXIMUM PERFORMANCE
DGMGRL>
Connect sys
Password:
DGMGRL> show configuration
Configuration
Name:
Enabled:
Protection Mode:
Fast-Start Failover:
pec
YES
MaxPerformance
DISABLED
Oracle 11g allows to implement Fast-Start Failover for the two modes “Maximum Availability” and also for
“Maximum Performance”.
Using “Maximal Performance” mode
As this is the default mode, no further changes are necessary. At all take care that LogXPTMode is still set to
'ASYNC'.
If it was set to 'ARCH' before, set it back to 'ASYNC'
DGMGRL> EDIT DATABASE o11_north SET PROPERTY LogXptMode='ASYNC';
DGMGRL> EDIT DATABASE o11_south SET PROPERTY LogXptMode='ASYNC';
Set the property “FastStartFailoverLagLimit”
DGMGRL> EDIT CONFIGURATION SET PROPERTY FastStartFailoverLagLimit=10;
This configuration property establishes an acceptable limit, in seconds, that the standby is allowed to fall
behind the primary in terms of redo applied, beyond which a fast-start failover will not be allowed. The lowest
possible value is 10 seconds.
This property is used when fast-start failover is enabled and the configuration is operating in maximum
performance mode.
Default value is set to 30 seconds.
Continue with Step 3.
Using “Maximal Availability” mode
First change the the matching Log transport mode (LogXptMode) to ‘SYNC’.
Configure redo transport services for this standby database using the LGWR, SYNC, and AFFIRM attributes
of the LOG_ARCHIVE_DEST_n initialization parameter. This mode, along with standby redo log files, is
required for the maximum protection or maximum availability protection modes. This redo transport service
enables the highest grade of data protection to the primary database, but also incurs the highest performance
impact.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
DGMGRL> EDIT DATABASE PEC_hpcoesd1 SET PROPERTY LogXptMode='SYNC';
DGMGRL> EDIT DATABASE PEC_hpcoe8640r2 SET PROPERTY LogXptMode='SYNC';
Then change the protection mode of configuration to Max Availability Protection.
This protection mode provides the highest level of data protection that is possible without compromising the
availability of the primary database. Like maximum protection mode, a transaction will not commit until the
redo needed to recover that transaction is written to the local online redo log and to at least one remote
standby redo log. Unlike maximum protection mode, the primary database does not shut down if a fault
prevents it from writing its redo stream to a remote standby redo log. Instead, the primary database operates
in maximum performance mode until the fault is corrected, and all gaps in redo log files are resolved. When
all gaps are resolved, the primary database automatically resumes operating in maximum availability mode.
This mode ensures that no data loss will occur if the primary database fails, but only if a second fault does not
prevent a complete set of redo data from being sent from the primary database to at least one standby
database.
This protection mode is required if you enable Fast-Start failover.
DGMGRL> EDIT CONFIGURATION SET PROTECTION MODE AS MAXAVAILABILITY;
DGMGRL> show configuration
Configuration - o11
Protection Mode: MaxAvailability
Databases:
o11_north - Primary database
o11_south - Physical standby database
Fast-Start Failover: DISABLED
SQLPLUS:
SQL> SELECT PROTECTION_MODE, PROTECTION_LEVEL FROM V$DATABASE;
PROTECTION_MODE
PROTECTION_LEVEL
-------------------- -------------------MAXIMUM AVAILABILITY MAXIMUM AVAILABILITY
It is not necessary to restart database after have changed the protection mode.
Step 3: Configure REAL TIME APPLY
Running real time apply requires to have standby redo logs in place. Setting up standy redo logs is allready
described in this whitepaper. To configure real time apply of redo stream, the broker property DelayMins is
set to no value.
EDIT DATABASE o11_north SET PROPERTY DelayMins='';
EDIT DATABASE o11_south SET PROPERTY DelayMins='';
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Step 4: Configure Fast-Start Failover
Specify the target standby database with the FastStartFailoverTarget property.
DGMGRL> EDIT DATABASE o11_north
SET PROPERTY FastStartFailoverTarget = 'o11_south';
DGMGRL> EDIT DATABASE o11_south
SET PROPERTY FastStartFailoverTarget = 'o11_north';
The target will be automated set to the standby database if only 1 standby database is configured. In any other
case it has to be set be the commands above.
Step 5: Set the FastStartFailoverThreshold property.
This property manage the time for failover. The default is set to 30 seconds, setting it to a different value
overwrites the default and will give DBA a longer period to possible stop the countdown.
DGMGRL> EDIT CONFIGURATION SET PROPERTY FastStartFailoverThreshold = 60;
This value depends on individual system level needs. It should be as short as possible but also give a realistic
chance to prevent the failover, if the primary database would be get back to business in time, may be due to a
short network outage, so a failover is no longer necessary.
Step 6: Setup an Observer
An Observer can be created at any database home or particulary installed administrative client, which has a
permanent network connection to the primary and the standby database.
For example, it can be setup at the consolidation system of a SAP landscape.
There is no need to install any different software if an available oracle home can be used. The observer will be
provided by DGMGRL as a forground process.
In the following example an observer is configured for an separate unix user, which has permissions to read
and execute on an existing oracle home from an different installation.
Example:
1.) create a user “observer” with group “dba”
2.) logon with user “observer”
3.) set environment to
ORACLE_BASE=/oracle
ORACLE_HOME=/oracle/<SID>/112_64
LD_LIBRARY_PATH=$ORACLE_HOME/lib
TNS_ADMIN=$HOME/config
PATH=$ORACLE_HOME:$PATH
4.) create directory $HOME/config
5.) copy sqlnet.ora and tnsnames.ora from your primary database home to $HOME/config
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
6.) test with tnsping if the primary and standby database are reachable.
7.) Start “dgmgrl” to test if the utility can be executed.
To start the observer, only the primary database must be running; it is not necessary for the target standby
database to be running.
You can start the observer before or after you enable fast-start failover. However, it is recommended that you
have the observer running whenever you have fast-start failover enabled. If Fast-Start failover is enabled, the
observer immediately begins monitoring the status and connections to the primary and target standby
databases.
To start the observer with DGMGRL, run the following command on the observer computer:
DGMGRL> connect sys@O11
Password: ********
DGMGRL> START OBSERVER;
This creates a foreground process, the prompt will never come back until a
DGMGRL> STOP OBSERVER
is executed from an second session. A scripted option which will log all output into an logfile and does not
need to have a permanet window open for observer is described later.
After the observer was started the first time, Fast-Start Failover can now be enabled at Data Guard
configuration.
Step 7: Enable Fast-Start Failover
To enable fast-start failover with DGMGRL, execute the ENABLE FAST_START FAILOVER command
while connected to any database system in the broker configuration (but not to the observer). For example:
DGMGRL> ENABLE FAST_START FAILOVER;
Enabled.
DGMGRL> show configuration verbose
Configuration - o11
Protection Mode: MaxPerformance
Databases:
o11_north - Primary database
o11_south - (*) Physical standby database
(*) Fast-Start Failover target
Fast-Start Failover: ENABLED
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Threshold:
Target:
Observer:
Lag Limit:
Shutdown Primary:
Auto-reinstate:
60 seconds
o11_south
oracle-vm2
10 seconds
TRUE
TRUE
Configuration Status:
SUCCESS
Step 8 Verfify the Fast_Start Failover environment
To verify the readiness of the fast-start failover configuration, issue the DGMGRL SHOW
CONFIGURATION VERBOSE command or the SHOW FAST_START FAILOVER command on the
primary database. For example:
DGMGRL> show fast_start failover
Fast-Start Failover: ENABLED
Threshold:
Target:
Observer:
Lag Limit:
Shutdown Primary:
Auto-reinstate:
60 seconds
o11_south
oracle-vm2
300 seconds
TRUE
TRUE
Configurable Failover Conditions
Health Conditions:
Corrupted Controlfile
Corrupted Dictionary
Inaccessible Logfile
Stuck Archiver
Datafile Offline
YES
YES
NO
NO
YES
Oracle Error Conditions:
(none)
Maintain the Observer
For generally information about working with an observer see Oracle documentation: Data Guard Broker
Managing the observer.
Start the observer process:
Dgmgrl> connect sys@O11
Password:*******
DGMGRL> START OBSERVER
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Using a logfile
As observer is running as a forground process, sometimes it would be better to run it in background and have
all output logged to a file. Use the example below to execute in background:
oracle-vm2> dgmgrl -logfile observer.log sys/<password>@O11 "start observer" &
A logfile “observer.log” will created into the current directory.
For the connect we use a full qualified user account including password. As this may be not a secure option,
because the SYS password is visible at the command line at particular operation systems, another way would
be more secure to use
dgmgrl -logfile observer.log /@edu "start observer" &
but this need Oracle Walet Manager (Oracle Database Advanced Security Administrator's Guide - section 9
Using Oracle Walet Manager ) to be set up.
Test is ps –efa | grep dgmgrl give a password line back at you operating system:
observer 6416 6247 0 15:37 pts/2
dgmgrl -logfile observer.log
00:00:00
start observer
if not, using a full qualified login should not be a problem.
Stop the observer:
Open a second window with the observer unix user and type:
Dgmgrl> connect sys@O11
Password:*******
DGMGRL> STOP OBSERVER
Or using a command line :
dgmgrl sys/manager@O11 "stop observer"
this can easy be scripted for a quicker use.
Monitoring:
You can check also at database if the Observer is present and the Fast-Start Failover status is still
synchronized:
SQL> Select FS_FAILOVER_STATUS, FS_FAILOVER_OBSERVER_PRESENT from v$database;
FS_FAILOVER_STATUS
FS_FAIL
--------------------- ------SYNCHRONIZED
YES
View the Data Guard Broker “Fast-Start Failover Configuration Statistics and Status” from documentation
for the value of column FS_FAILOVER_STATUS.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Disable Fast-Start Failover
There may be some reason to disable the feature. An overview can be find at section 5.5.4 in Oracle Data
Guard Broker documentation. Disabling this let you still perform manual failover or switchover your Data
Guard configuration.
The observer will not stopped by disabling Fast-start failover, it need stopped manually.
To disable Fast-Start failover in DGMGRL use:
DGMGRL> DISABLE FAST-START FAILOVER
[FORCE];
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Reconnect SAP instance to database
In principle, some different approaches are possible,
•
using a virtual host name / IP address as is standard with a cluster – means: no changes in SAP profiles
and SQL*Net config files in case of desaster
•
using Client connection timeout with DATABASE_ROLE change event to trigger the database
service start at the new primary side. - means: no changes in SAP profiles and SQL*Net config files in
case of desaster
•
using prepared and tested profile files with which the CI and application servers can be restarted
should a disaster occur. - means: no changes at the SQL*Net profiles, but at SAP profiles in case of a
desaster.
In all cases the SAP CI must again be able to access the database using the parameters set in the profile files.
The CI must either be operated on a system independent of the database server, which would therefore still
be available after the primary database server had experienced the disaster, or must then be restarted on a
failover server. The failover server can also be identical to the secondary database server.
Consider to use one of the first both topics to have a gracefull reconnect of SAP without to have to restart
any of that components. See the below SQL*Net configuration for details.
Reconnecting the CI
The SAP instance can either be reconnected by the newly started CI, which is on-hand with the previously
preconfigured profile file DEFAULT.PFL and only actually starts to operate in the event of a disaster.
Set the following SAP profile parameters in the transaction RZ20 on the disaster CI or manually at file level at
the default profile of SAP instance.
The file DEFAULT.PFL can be found in directory /usr/SAP/PEC/SYS/profile.
Set parameter SAPDBHOST to the current database hostname after the switch:
SAPDBHOST = <hostname of Standby DB>
There is no need to change the SID, as it will be still the same after a database role switch.
After that change start SAP CI to become parameter change active.
Each SAP instance has a start profile which possible includes an entry for setting the environment variable
“dbs_ora_tnsname=” defining the database service for connect. Such a service will be provided by listsner
and has to be configurat as a Net Service Name in tnsnames.ora.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Alternatively, if a virtual IP / host name is available or Client connection timeout setup is used, the disaster CI
can run with the same parameter settings as the original CI. The modified host name is realized using the
network layer. SAP CI is still bound to the previously defined VIP host name, even in case of a independent
running SAP CI reconnection feature will take place.
SAP CI will try reconnect to database until database service will become available again.
Reconnecting the application server
The profile parameter for the message server must be adapted to the application servers.
However, if a virtual host name is used, no changes are required. The application servers must be restarted in
order to log back onto the new CI (message server).
Enter the new message server on the application servers instance profiles and restart the instances.
rdisp/msname = <Host name of disaster CI>
SQL*Net configuration
Provided a virtual host name is not used, consider to use the decribed solution for the Client Connection
timeout. Otherwise configuration in the TNSNAMES.ORA must be adapted to the disaster CI.
We would recommend entering and testing this in advance. If the disaster CI is on the same host as the
standby database, the TNSNAMES.ORA of the database installation can be used if, for example, the
environment variable TNS_ADMIN refers to the database home directory. TNS_ADMIN has been used
again in combination with the instant client since 10g.
Client connection Timeout
This feature is available with 10.2 databases and above. It is based on load balancing entries and connect
timeout settings in SQL*Net. A very good and detailed description for setting up this configuration can be
found in the Oracle MAA Whitepaper “Client Failover Best Practices for Highly Available Oracle Databases:
Oracle Database 10g Release 2”. Additional it needs to setup a database trigger for switching the database
service at database role change event.
Example:
In TNSNAMES.ORA of client side, at SAP generally found at : /usr/sap/<SID>/SYS/profile/oracle
Create an Oracle Net service name that includes all primary and standby hosts in the ADDRESS_LIST, and
allow load balancing.
O11_DVEBMGS00 =
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL = TCP)(HOST = oracle-lnx)(PORT = 1527))
(ADDRESS = (PROTOCOL = TCP)(HOST = oracle-vm2)(PORT = 1527))
(LOAD_BALANCE = yes)
)
(CONNECT_DATA=
(SERVICE_NAME=o11)
)
)
In the client side SQLNET.ORA file, set the SQLNET.OUTBOUND_CONNECT_TIMEOUT parameter.
This parameter enables clients to quickly traverse an address_list in the event of a failure. For example, if a
client attempts to connect to a host that is unavailable, the connection attempt will be bounded to the time
specified by the SQLNET.OUTBOUND_CONNECT_TIMEOUT parameter, after which the client
attempts to connect to the next host in the address_list. This behavior continues for each host in the
address_list until a connection is made. Setting the parameter to a value of 3 seconds will suffice in most
environments.
SQLNET.OUTBOUND_CONNECT_TIMEOUT= t (time in seconds).
Switching database service at database role event
To make this working the database needs to provide a service “<SID>”, in our example it would be “O11”.
Whereby we can possible connect to one of the above listed databases, as we will done using normal
loadbalancing, we will find only 1 active primary database, in our Data Guard configuration which provide
this service at one time.
May be the service is provided by the original primary instance or - after a desaster - by the overtaking standby
database, which now owns the database role “PRIMARY” .
Switching this service is based an a database trigger, which will only fire at startup of database instance if the
database role is “Primary”.
Such a service is set as default to SID for each the datasbase.
All the created services can be listet using the query:
SQL> select service_id, name, enabled from dba_services;
SERVICE_ID
---------1
2
3
NAME
---------------------------------------------------------------SYS$BACKGROUND
SYS$USERS
O11
ENA
--NO
NO
NO
This service is disabled by default, but can be enabled by a simple PL/SQL.
Another way would be, to define such a service in listener.ora. But that means, we have to modify or restart
the listener service after a desaster at the standby host to provide this service. Having service enabled at both
listeners at the same time, as we will get with a statically listener setup, will let fail connection entries, because
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
we cannot connect to a standby database with application users. Thus, this service should only be provided by
database, which is open for use and that is the one with database role “PRIMARY”.
Using DBMS_SERVICE.START_SERVICE enable the DBA to switch the service to the active database
node using scripted solutions like an “AFTER STARTUP” database trigger.
We have to create a database trigger that fires on the database startup event to relocate the database service
'O11' to a Data Guard Standby database after it has transitioned to the primary role.
Example:
CREATE OR REPLACE TRIGGER manage_DBservice
after startup on database
DECLARE
role VARCHAR(30);
BEGIN
SELECT DATABASE_ROLE INTO role FROM V$DATABASE;
IF role = 'PRIMARY' THEN
DBMS_SERVICE.START_SERVICE('O11');
END IF;
END;
/
checking with :
[oracle@oracle-vm2 dbs]$ lsnrctl services listener_o11
LSNRCTL for Linux: Version 11.2.0.1.0 - Production on 21-JAN-2010 11:41:53
Copyright (c) 1991, 2009, Oracle.
All rights reserved.
Connecting to (ADDRESS=(PROTOCOL=IPC)(KEY=O11.WORLD))
Services Summary...
Service "O11.WORLD" has 1 instance(s).
Instance "O11", status READY, has 1 handler(s) for this service...
Handler(s):
"DEDICATED" established:0 refused:0 state:ready
LOCAL SERVER
This trigger will also be installed at standby side due to the redo transport. Next time standby side become
primary it will fire and start the service.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Performance
Various things may cause the log apply to not perform on the standby side and the database to continue to
relapse in the recovery over time. This means that in the event of a disaster, the specified times (SLA) cannot
be observed because the database cannot be recovered fast enough.
It is therefore well worth monitoring the performance of the standby database even during normal operation
and intervening as early on as possible. Various V$ views are available for this in mount status. Normal
monitoring, e.g. using AWR views, is not possible because the database has to be open for this.
Performance monitoring
The most important performance views can be queried using SQLPLUS. One very common approach is
provided by the Oracle Maximum Availability Architecture (MAA) paper "DataGuard Redo Apply and Media
Recovery best Practices", which can be found at:
http://www.oracle.com/technology/deploy/availability/pdf/MAA_WP_10gR2_RecoveryBestPractices.pdf.
Parallel recovery
Another way of achieving a good recovery performance is to use parallel recovery.
This approach is however only worthwhile on servers with at least 8 CPUs because the overhead for
managing the parallel processes otherwise partially cancels out the gain in performance.
When using Data Guard Broker, parallel recovery can be controlled using the "ApplyParallel" property.
DGMGRL> EDIT DATABASE <name> SET PROPERTY ApplyParallel='AUTO'
Possible values are:
•
AUTO: The number of processes for the "Redo Apply" is calculated automatically by Oracle on the basis
of the CPUs available. This is the default setting.
•
NO: Parallel recovery is deactivated.
The following parameters should be set on the standby database. After a role transition, they should also be
set on the now active standby database:
PARALLEL_MAX_SERVERS = <CPU*2>
DB_BLOCK_CHECKING=FALSE
PARALLEL_EXECUTION_MESSAGE >= 16348
(max = 65536)
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Use a test run to compare the performance of serial and parallel recovery to establish the actual gain in
performance.
As the number of parallel processes increases, CPU utilization and the I/O bandwidth used also goes up.
Parallel recovery and the recovery slave processes use more system resources because of the IPC messaging
overhead and recovery communication. Excess overheads may mean that the gain in performance is canceled
out.
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Annex
Database parameters of relevance to Data Guard, primary database
*.db_unique_name='o11_north'
*.dg_broker_start=true
*.local_listener='(ADDRESS = (PROTOCOL = TCP)(HOST = oracle-lnx)(PORT = 1527))'
*.log_archive_dest_1='LOCATION="/oracle/O11/oraarch/O11arch",
valid_for=(ONLINE_LOGFILE,ALL_ROLES)'
*.log_archive_dest_state_1='ENABLE'
*.log_archive_format='%t_%s_%r.dbf'
*.log_archive_max_processes=2
*.log_archive_min_succeed_dest=1
*.log_archive_trace=0
*.log_file_name_convert='O11','O11'
*.standby_file_management='AUTO'
Database parameters of relevance to Data Guard, standby database
*.db_unique_name='o11_south'
*.dg_broker_start=TRUE
*.local_listener='(ADDRESS = (PROTOCOL = TCP)(HOST = oracle-vm2)(PORT = 1527))'
*.log_archive_dest_1='location="/oracle/O11/oraarch/O11arch" mandatory
valid_for=(online_logfile,all_roles)'
*.log_archive_dest_state_1='ENABLE'
*.log_archive_format='%t_%s_%r.dbf'
*.log_archive_max_processes=2
*.log_archive_min_succeed_dest=1
*.log_archive_trace=0
*.log_file_name_convert='O11','O11'
*.standby_file_management='AUTO'
Listener settings, primary database
ADMIN_RESTRICTIONS_LISTENER_O11 = on
LISTENER_O11 =
(ADDRESS_LIST =
(ADDRESS =
(PROTOCOL = IPC)
(KEY = O11.WORLD)
)
(ADDRESS=
(PROTOCOL = IPC)
(KEY = O11)
)
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-lnx)
(PORT = 1527)
)
)
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
STARTUP_WAIT_TIME_LISTENER_O11 = 0
CONNECT_TIMEOUT_LISTENER_O11 = 10
TRACE_LEVEL_LISTENER_O11= OFF
SID_LIST_LISTENER_O11=
(SID_LIST =
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-lnx)
(ORACLE_HOME = /oracle/O11/1102_64)
)
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-lnx_DGMGRL)
(ORACLE_HOME = /oracle/O11/1102_64)
)
# optional
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-lnx_DGMGRL.WORLD)
(ORACLE_HOME = /oracle/O11/1102_64)
)
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-lnx_XPT.WORLD)
(ORACLE_HOME = /oracle/O11/1102_64)
)
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-lnx_DGB.WORLD)
(ORACLE_HOME = /oracle/O11/1102_64)
)
# end optional
)
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Listener settings, standby database
ADMIN_RESTRICTIONS_LISTENER_O11 = on
LISTENER_O11 =
(ADDRESS_LIST =
(ADDRESS =
(PROTOCOL = IPC)
(KEY = O11.WORLD)
)
(ADDRESS=
(PROTOCOL = IPC)
(KEY = O11)
)
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-vm2)
(PORT = 1527)
)
)
STARTUP_WAIT_TIME_LISTENER_O11 = 0
CONNECT_TIMEOUT_LISTENER_O11 = 10
TRACE_LEVEL_LISTENER_O11= OFF
SID_LIST_LISTENER_O11=
(SID_LIST =
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-vm2)
(ORACLE_HOME = /oracle/O11/1102_64)
)
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-vm2_DGMGRL)
(ORACLE_HOME = /oracle/O11/1102_64)
)
# optional
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-vm2_DGMGRL.WORLD)
(ORACLE_HOME = /oracle/O11/1102_64)
)
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-vm2_XPT.WORLD)
(ORACLE_HOME = /oracle/O11/1102_64)
)
(SID_DESC =
(SID_NAME = O11)
(GLOBAL_DBNAME=o11_oracle-vm2_DGB.WORLD)
(ORACLE_HOME = /oracle/O11/1102_64)
)
# end optional
)
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Tnsnames.ora, primary database
O11_oracle-lnx.WORLD=
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-lnx)
(PORT = 1527)
)
)
(CONNECT_DATA =
(SID = O11)
(GLOBAL_NAME = O11)
(SERVICE_NAME = o11_oracle-lnx)
)
)
O11_oracle-vm2.WORLD=
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-vm2)
(PORT = 1527)
)
)
(CONNECT_DATA =
(SID = O11)
(SERVICE_NAME= o11_oracle-vm2)
)
)
Tnsnames.ora, standby database
O11_oracle-vm2.WORLD=
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-vm2)
(PORT = 1527)
)
)
(CONNECT_DATA =
(SID = O11)
(GLOBAL_NAME = O11)
(SERVICE_NAME = o11_oracle-vm2)
)
)
O11_oracle-lnx.WORLD=
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS =
(COMMUNITY = SAP.WORLD)
(PROTOCOL = TCP)
(HOST = oracle-lnx)
(PORT = 1527)
)
)
(CONNECT_DATA =
(SID = O11)
(SERVICE_NAME = o11_oracle-lnx)
)
)
Script: cre_stdbycontrolfile.sql
alter database create standby controlfile as
'/oracle/O11/sapreorg/stdby_controlf.dbf' reuse;
scp -r /oracle/O11/sapreorg/stdby_controlf.dbf
vm2:/oracle/O11/sapreorg
oracle@oracle-
Script: cpy_controlf.sh
cp /oracle/O11/sapreorg/stdby_controlf.dbf /oracle/O11/origlogA/cntrl/cntrlO11.dbf
cp /oracle/O11/sapreorg/stdby_controlf.dbf /oracle/O11/origlogB/cntrl/cntrlO11.dbf
cp /oracle/O11/sapreorg/stdby_controlf.dbf /oracle/O11/sapdata1/cntrl/cntrlO11.dbf
Creating the Data Guard configuration
create configuration o11 as primary database is o11_north connect identifier is
o11_oracle-lnx;
Add physical standby database to DG config.
add database o11_south as connect identifier is o11_oracle-vm2 maintained as
physical;
Oracle White Paper—Setting up Oracle 11g Data Guard for SAP customers
Data Guard properties
EDIT DATABASE o11_north SET PROPERTY LogXptMode='ASYNC';
EDIT DATABASE o11_north SET PROPERTY DelayMins='5';
EDIT DATABASE o11_north SET PROPERTY MaxFailure='100';
EDIT DATABASE o11_north set PROPERTY logfilenameconvert ="'O11','O11'";
EDIT DATABASE o11_north set PROPERTY hostname = 'oracle-lnx';
EDIT DATABASE o11_north set PROPERTY
StandbyArchiveLocation='/oracle/O11/oraarch/O11arch'
EDIT DATABASE o11_south SET PROPERTY LogXptMode='ASYNC';
EDIT DATABASE o11_south SET PROPERTY DelayMins='5';
EDIT DATABASE o11_south SET PROPERTY MaxFailure='100';
EDIT DATABASE o11_south set PROPERTY hostname = 'oracle-vm2';
EDIT DATABASE o11_south set PROPERTY logfilenameconvert ="'O11','O11'";
EDIT DATABASE o11_south set PROPERTY
StandbyArchiveLocation='/oracle/O11/oraarch/O11arch'
Standby log files
mkdir /oracle/O11/standbylog
alter database
reuse;
alter database
reuse;
alter database
reuse;
alter database
reuse;
alter database
reuse;
add standby logfile '/oracle/O11/standbylog/srl1.dbf' size 100M
add standby logfile '/oracle/O11/standbylog/srl2.dbf' size 100M
add standby logfile '/oracle/O11/standbylog/srl3.dbf' size 100M
add standby logfile '/oracle/O11/standbylog/srl4.dbf' size 100M
add standby logfile '/oracle/O11/standbylog/srl5.dbf' size 100M
References
Data Guard Broker
Data Guard Concepts and Administration
Oracle Database High Availability Best Practices 10g Release 2 Documentation
Setup Flashback Database on Data Guard Physical Standby Database for SAP
Customers
HTML PDF
HTML PDF
HTML
HTML
Setting up Oracle 11g Data Guard
for SAP customers
January 2010
Author: Bernhard Baumann
Oracle Support Solution Center SAP,
Walldorf (Germany)
Contributing Authors: Some of the contents
of this document is taken from Oracle
Copyright © 2009, Oracle and/or its affiliates. All rights reserved. This document is provided for information purposes only and
documentation and also from Oracle
the contents hereof are subject to change without notice. This document is not warranted to be error-free, nor subject to any other
White Paper “Oracle Data Guard in Oracle
warranties or conditions, whether expressed orally or implied in law, including implied warranties and conditions of merchantability or
Database 10g Release 2” written by
fitness for a particular purpose. We specifically disclaim any liability with respect to this document and no contractual obligations are
Ashish Ray in 2006
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.
Oracle Corporation
World Headquarters
500 Oracle Parkway
Redwood Shores, CA 94065
Oracle is a registered trademark of Oracle Corporation and/or its affiliates. Other names may be trademarks of their respective
owners.
U.S.A.
1009
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

advertising