4 Managing Outages

4.1.1.1 Managing Unscheduled Outages on the Primary Site

If the primary site contains the production database and the secondary site contains the standby database, then outages on the primary site are the most crucial. Solutions for these outages are critical for maximum availability of the system. Only the Oracle Database 10g with Data Guard, and the Oracle Database 10g with RAC and Data Guard (MAA) architectures have a secondary site to protect from site disasters.

Table 4-2 summarizes the recovery steps for unscheduled outages on the primary site. For outages that require multiple recovery steps, the table includes links to the detailed descriptions in Section 4.2, "Recovering from Unscheduled Outages" that starts.

^Footnote 1 Recovery time indicated applies to database and existing connection failover. Network connection changes and other site-specific failover activities may lengthen overall recovery time.

^Footnote 2 Recovery time consists largely of the time it takes to restore the failed system.

^Footnote 3 Database is still available, but portion of application connected to failed system is temporarily affected.

^Footnote 4 Storage failures are prevented by using Automatic Storage Management (ASM) with mirroring and its automatic rebalance capability.

^Footnote 5 Not all types of data corruption are prevented. For the most recent information about the HARD initiative, refer to Section 2.1.6, "Consider HARD-Compliant Storage".

^Footnote 6 Recovery times from human errors depend primarily on detection time. If it takes seconds to detect a malicious DML or DLL transaction, then it typically only requires seconds to flashback the appropriate transactions.

4.1.1.2 Managing Unscheduled Outages on the Secondary Site

For most cases, outages on the secondary site can be managed with no effect on availability of the primary database located on the primary site. However, if the configuration is in maximum protection mode, then unscheduled outages on the last surviving standby database will cause outages on the production database to ensure no data loss when failing over to the standby database. After downgrading the data protection mode, you can restart the production database even without accessibility to the standby databases. Outages on the secondary site might affect the maximum time to recovery (MTTR) if there are concurrent failures on the primary site.

Table 4-3 summarizes the recovery steps for unscheduled outages of the standby database on the secondary site. For outages that require multiple recovery steps, the table includes links to the detailed descriptions in Section 4.2, "Recovering from Unscheduled Outages" that starts.

4.1.2.1 Managing Scheduled Outages on the Primary Site

If the primary site contains the production database and the secondary site contains the standby database, then outages on the primary site are the most crucial. Solutions for theses outages are critical for continued availability of the system.

Table 4-5 shows the high-level recovery steps for scheduled outages on the primary site. For outages that require multiple recovery steps, the table includes links to the detailed descriptions in Section 4.4, "Eliminating or Reducing Downtime for Scheduled Outages" beginning.

4.1.2.2 Managing Scheduled Outages on the Secondary Site

Outages on the secondary site do not affect availability because the clients always access the primary site. Outages on the secondary site might affect the RTO if there are concurrent failures on the primary site. Outages on the secondary site can be managed with no effect on availability. If maximum protection database mode is configured, then downgrade the protection mode before scheduled outages on the standby instance or database so that there will be no downtime on the production database.

Table 4-6 describes the recovery steps for scheduled outages on the secondary site.

4.1.2.3 Preparing for Scheduled Outages on the Secondary Site

To achieve continued service during scheduled outages on a secondary site when in maximum protection mode, downgrade the maximum protection mode to maximum availability or maximum performance temporarily. When scheduling secondary site maintenance, consider that the duration of a site-wide or clusterwide outage adds to the time that the standby database lags behind the production database, which in turn lengthens the time to restore fault tolerance. See Section 2.4.2, "Data Protection Mode" for an overview of the Data Guard protection modes.

4.2.1.1 When to Use Complete Site Failover

If the standby site meets the prerequisites, then complete site failover is recommended for the following scenarios:

• Primary-site disaster, such as natural disasters or malicious attacks

• Primary network-connectivity failures

• Primary site power failures

4.2.1.2 Best Practices for Complete Site Failover

Site failover can be expedited in minutes by using the following practices:

• Use Data Guard configuration best practices

• Use Data Guard fast-start failover to automatically fail over to the standby database, with a recovery time objective (RTO) of less than 30 seconds

• Maintain a running middle-tier application server on the secondary site to avoid the startup time

• Automate the DNS failover procedure

Data loss is dependent on the Data Guard configuration and the use of synchronous or asynchronous redo shipping.

4.2.1.3 Repair Solution

A wide-area traffic manager on the primary and secondary sites provides the site failover function. The wide-area traffic manager can redirect traffic automatically if the primary site, or a specific application on the primary site, is not accessible. It can also be triggered manually to switch to the secondary site for switchovers. Traffic is directed to the secondary site only when the primary site cannot provide service due to an outage or after a switchover. If the primary site fails, then user traffic is directed to the secondary site automatically.

Figure 4-1 illustrates the possible network routes before site failover:

1. Client requests enter the client tier of the primary site and travel by the WAN traffic manager.

2. Client requests are sent through the firewall into the demilitarized zone (DMZ) to the application server tier.

3. Requests are forwarded through the active load balancer to the application servers.

4. Requests are sent through another firewall and into the database server tier.

5. The application requests, if required, are routed to a RAC instance.

6. Responses are sent back to the application and clients by a similar path.

Figure 4-1 Network Routes Before Site Failover

Description of "Figure 4-1 Network Routes Before Site Failover"

Figure 4-2 illustrates the network routes after site failover. Client or application requests enter the secondary site at the client tier and follow exactly the same path on the secondary site that they followed on the primary site.

Figure 4-2 Network Routes After Site Failover

Description of "Figure 4-2 Network Routes After Site Failover"

The following steps describe the effect on network traffic of a failover or switchover:

1. The administrator has failed over or switched over the production database to the secondary site. This is automatic if you are using Data Guard fast-start failover.

2. The administrator starts the middle-tier application servers on the secondary site, if they are not already running.

3. The wide-area traffic manager selection of the secondary site can be automatic in the case of an entire site failure. The wide-area traffic manager at the secondary site returns the virtual IP address of a load balancer at the secondary site and clients are directed automatically on the subsequent reconnect. In this scenario, the site failover is accomplished by an automatic domain name system (DNS) failover.

   Alternatively, a DNS administrator can manually change the wide-area traffic manager selection to the secondary site for the entire site or for specific applications. The following is an example of a manual DNS failover:

   1. Change the DNS to point to the secondary site load balancer:

      The master (primary) DNS server is updated with the new zone information, and the change is announced with DNS NOTIFY.

      The slave DNS servers are notified of the zone update with a DNS NOTIFY announcement, and the slave DNS servers pull the new zone information.

      Note:

      The master and slave servers are authoritative name servers. Therefore, they contain trusted DNS information.

   2. Clear affected records from caching DNS servers.

      A caching DNS server is used primarily for performance and fast response. The caching server obtains information from an authoritative DNS server in response to a host query and then saves (caches) the data locally. On a second or subsequent request for the same data, the caching DNS server responds with its locally stored data (the cache) until the time-to-live (TTL) value of the response expires. At this time, the server refreshes the data from the zone master. If the DNS record is changed on the primary DNS server, then the caching DNS server does not pick up the change for cached records until TTL expires. Flushing the cache forces the caching DNS server to go to an authoritative DNS server again for the updated DNS information.

      Flush the cache if the DNS server being used supports such a capability. The following is the flush capability of common DNS BIND versions:

      BIND 9.3.0: The command rndc flushname name flushes individual entries from the cache.

      BIND 9.2.0 and 9.2.1: The entire cache can be flushed with the command rndc flush.

      BIND 8 and BIND 9 up to 9.1.3: Restarting the named server clears the cache.

   3. Refresh local DNS service caching.

      Some operating systems might cache DNS information locally in the local name service cache. If so, this cache must also be cleared so that DNS updates are recognized quickly.

      Solaris: nscd

      Linux: /etc/init.d/nscd restart

      Microsoft Windows: ipconfig /flushdns

      Apple Mac OS X: lookupd -flushcache

   4. The secondary site load balancer directs traffic to the secondary site middle-tier application server.

   5. The secondary site is ready to take client requests.

Failover also depends on the client's Web browser. Most browser applications cache the DNS entry for a period of time. Consequently, sessions in progress during an outage might not fail over until the cache timeout expires. To resume service to such clients, close the browser and restart it.

4.2.2.1 When To Perform a Data Guard Failover

When a primary database failure cannot be repaired in time to meet your Recovery Time Objective (RTO) using local backups or Flashback Technology, Data Guard should be used.

A manual failover, which is user initiated, should be performed due to an unplanned outage such as:

• Site disaster which results in the primary database becoming unavailable

• User errors that cannot be repaired in a timely fashion

• Data failures, which impact the production application

A failover requires that the initial production database be reinstated as a standby database to restore fault tolerance to your environment. The standby database can be quickly reinstated by using Flashback Database. See Section 4.3.2, "Restoring a Standby Database After a Failover".

4.2.2.2 Best Practices for Implementing Data Guard Failover

A fast-start failover is completely automated and requires no user intervention. A manual failover, being user-driven, can be performed using Enterprise Manager, the Data Guard broker command-line interface, or SQL*Plus commands:

• Fast-start failover: There are no procedural best practices to consider when performing a fast-start failover. However, it is very important to address all of the configuration best practices described in Section 2.4.7.2.2, "Fast-Start Failover Best Practices".

• Manual failover: When performing a manual failover, follow the best practices described in Section 4.2.2.2.2, "Using SQL to Fail Over to a Physical Standby Database" and the configuration best practices outlined in Section 2.4.7.2.3, "Manual Failover Best Practices":

  For manual failovers that involve Real Application Clusters, issue the SHUTDOWN ABORT statement on all secondary RAC instances on the standby database prior to performing a failover.

This section contains these topics:

• Using Enterprise Manager to Perform a Data Guard Failover

• Using SQL to Fail Over to a Physical Standby Database

• Using SQL to Fail Over to a Logical Standby Database

4.2.2.2.1 Using Enterprise Manager to Perform a Data Guard Failover

The procedure for Data Guard failover is the same for both physical and logical standby databases. The following screen shots illustrate how to perform a failover using Oracle Enterprise Manager:

In Figure 4-3 the Data Guard Overview page shows the ORA-16625 error status that indicates problems accessing the primary database.

Figure 4-3 Data Guard Overview Page Showing ORA-16625 Error

Description of "Figure 4-3 Data Guard Overview Page Showing ORA-16625 Error"

To transition DR_Sales into the primary role, select DR_Sales in the Standby Databases table and click Failover.

Figure 4-4 shows the Failover Confirmation page.

Figure 4-4 Failover Confirmation Page

Description of "Figure 4-4 Failover Confirmation Page"

If you determine that a failure occurred on the primary database and there is no possibility of recovering the primary database in a timely manner, you can start the Failover operation. In configurations with both physical and logical standby databases, Oracle recommends using the physical standby database as the failover target because it will allow the logical standby database to continue to function as a logical standby to the new primary database. If the failover is made to the logical standby, any physical standbys in the configuration will need to be re-created from a backup of the new primary database.

The failover operation enables you to choose one of the following two types of failover operations:

• Complete

  This operation attempts to minimize data loss by applying all available redo on the standby database.

• Immediate

  No additional data is applied on the standby database; data might be lost. This is the fastest type of failover.

Figure 4-5 shows the progress of the failover operation.

Figure 4-5 Failover Progress Page

Description of "Figure 4-5 Failover Progress Page"

During the failover, the selected standby database (also referred to as the target standby database) transitions into the primary role. If the failover target is a physical standby database, it is restarted. When completed, the Data Guard Overview page reflects the updated configuration, as shown in Figure 4-6.

Figure 4-6 Data Guard Overview Page After a Failover Completes

Description of "Figure 4-6 Data Guard Overview Page After a Failover Completes"

In the figure, the Data Guard Status column indicates that the original primary database (North_Sales) is disabled and can no longer be managed through Enterprise Manager until it has been re-enabled as a physical standby database.

4.2.3.1 When to Perform a Data Guard Switchover

Switchover is a planned operation. Switchover is the capability to switch database roles between the production and standby databases. Switchover can occur whenever a production database is started, the target standby database is available, and all the archived redo logs are available. Switchovers are typically completed in less than five minutes and, in some cases, are optimized to be less than a minute. Switchovers are useful in the following situations:

• Scheduled maintenance such as hardware maintenance or firmware patches on the production host

• Resolution of data failures when the production database is still opened

• Testing and validating the secondary resources, as a means to test disaster recovery readiness

Switchover is not possible or practical under the following circumstances:

• Missing archived redo logs

• Point-in-time recovery is required

• The production database is not open and cannot be opened

4.2.3.2 Best Practices for Implementing Data Guard Switchover

Before performing a switchover the following procedural best practices should be followed in addition to the configuration best practices outlined in Section 2.4.7.1.1, "Switchover Best Practices":

1. Disconnect all sessions possible by issuing SQL command ALTER SYSTEM KILL SESSION.

2. Stop job processing by setting the AQ_TM_PROCESSES to 0.

3. Cancel any specified apply delay by using the NODELAY keyword to stop and restart log apply services on the standby database:

   • On a physical standby database:

     ALTER DATABASE RECOVER MANAGED STANDBY DATABASE CANCEL;
     ALTER DATABASE RECOVER MANAGED STANDBY DATABASE USING CURRENT LOGFILE NODELAY;
     
     

   • On a logical standby database:

     ALTER DATABASE STOP LOGICAL STANDBY APPLY;
     ALTER DATABASE START LOGICAL STANDBY APPLY IMMEDIATE NODELAY;
     
     

This following sections describe how to perform a switchover:

• Using Enterprise Manager to Perform a Data Guard Switchover

• Using SQL for Data Guard Switchover to a Physical Standby Database

• Using SQL for Data Guard Switchover to a Logical Standby Database

4.2.3.2.1 Using Enterprise Manager to Perform a Data Guard Switchover

The procedure for Data Guard switchover using Enterprise Manager is the same for both physical and logical standby databases:

1. Select the standby database that you want to become the primary database.

2. Click Switchover.

3. Click Yes to continue with the switchover. Click No to cancel.

Figure 4-7 shows the switchover confirmation page.

Figure 4-7 Switchover Operation Confirmation

Description of "Figure 4-7 Switchover Operation Confirmation"

Figure 4-8 shows the processing page during the switchover.

Figure 4-8 Processing Page During Switchover

Description of "Figure 4-8 Processing Page During Switchover"

Figure 4-9 shows the Data Guard overview page after a successful switchover.

Figure 4-9 New Primary Database After Switchover

Description of "Figure 4-9 New Primary Database After Switchover"

4.2.4.1 Automatic Instance Recovery for Failed Instances

Instance failure occurs when software or hardware problems disable an instance. After instance failure, Oracle automatically uses the online redo log file to perform database recovery as described in this section.

Instance recovery in RAC does not include restarting the failed instance or the recovery of applications that were running on the failed instance. Applications that were running continue by using service relocation and fast application notification (as described in Section 4.2.4.2, "Automatic Service Relocation").

When one instance performs recovery for another instance, the recovering instance:

• Reads redo log entries generated by the failed instance and uses that information to ensure that committed transactions are recorded in the database. Thus, data from committed transactions is not lost

• Rolls back uncommitted transactions that were active at the time of the failure and releases resources used by those transactions

When multiple node failures occur, as long as one instance survives, RAC performs instance recovery for any other instances that fail. If all instances of a RAC database fail, then on subsequent restart of any one instance, crash recovery will occur and all committed transactions will be recovered. If Data Guard is available, you can fail over automatically with Data Guard fast-start failover once all instances are down.

4.2.4.2 Automatic Service Relocation

Service reliability is achieved by configuring and failing over among redundant instances. More instances are enabled to provide a service than would otherwise be needed. If a hardware failure occurs and adversely affects a RAC database instance, then CRS automatically moves any services on that instance to another available instance, as configured with DBCA or Enterprise Manager. Then, Cluster Ready Services (CRS) attempt to restart the failed nodes and instances.

CRS recognizes when a failure affects a service and automatically fails over the service and redistributes the clients across the surviving instance supporting the service. In parallel, CRS attempts to restart and integrate the failed instances and dependent resources back into the system. Notification of failures using fast application notification (FAN) events occur at various levels within the Oracle Server architecture. The response can include notifying external parties through Oracle Notification Service (ONS), advanced queueing, or FAN callouts, recording the fault for tracking, event logging, and interrupting applications. Notification occurs from a surviving node when the failed node is out of service. The location and number of nodes serving a service is transparent to applications. Auto restart and recovery are automatic, including all the subsystems, such as the listener and the ASM instance, not just database.

4.2.4.3 Oracle Cluster Registry

Loss of the Oracle Cluster Registry file affects the availability of RAC and Cluster Ready Services. The OCR file can be restored from a physical backup that is automatically created or from an export file that is manually created by using the ocrconfig tool. Additionally, starting with Oracle Database 10g Release 10.2, Oracle can optionally mirror the OCR so that a single OCR device failure can be tolerated.

4.2.6.1 ASM Instance Failure

If the ASM instance fails, database instances accessing ASM storage from the same node will shut down.If the primary database is using RAC, then application failover will occur automatically and clients connected to the database instance will reconnect to remaining instances providing the required service in the cluster and continue processing. The recovery time is typically in seconds.

If the primary database does not use RAC, then an ASM instance failure will shut down the entire database. If Data Guard is being used and Data Guard fast-start failover is configured, a database failover will be triggered automatically and clients will automatically reconnect to the new primary database following the failover. The recovery time is the time it takes to complete an automatic Data Guard fast-start failover operation. If fast-start failover is not configured, then recovering from this outage is a manual process, which can be accomplished by either restarting the ASM and database instances manually, or by performing a Data Guard failover. If neither RAC nor Data Guard is being used, then restart the ASM instance and restart database instances manually. The recovery time depends on the length of time to start the ASM instance, and the length of time to start the database instances and perform crash recovery.

4.2.6.2 ASM Disk Failure

If an ASM disk group is configured as an external redundancy type, then a failure of a single disk should be handled by the storage array and should not be seen by the ASM instance, and all operations of ASM and databases using the disk group will continue normally. However, if the failure of an external redundancy disk group is seen by the ASM instance, then the ASM instance will take the disk group offline immediately, causing Oracle instances accessing the disk group to crash. If the disk failure is temporary, then ASM and the database instances can be restarted, and crash recovery will occur after the disk group is brought back online. If an ASM disk group is configured as a normal or a high-redundancy type, then disk failure is handled transparently by ASM and the databases accessing the disk group are not affected. An ASM instance automatically starts an ASM rebalance operation to distribute the data on one or more failed disks to alternative disks in the ASM disk group. While the rebalance operation is in progress, subsequent disk failures may affect disk group availability if the disk contains data that has yet to be re-mirrored. When the rebalance operation completes successfully, the ASM disk group is no longer at risk in the event of a subsequent failure. Multiple disk failures are handled similarly, provided the failures affect only one failure group in an ASM disk group.The failure of multiple disks in multiple failure groups where a primary extent and all of its mirror extents have been lost will cause the disk group to go offline.

This following recovery methods can be used:

• Using Enterprise Manager to Repair ASM Disk Failure

• Using SQL to Add Replacement Disks Back to the Disk Group

4.2.6.2.1 Using Enterprise Manager to Repair ASM Disk Failure

Figure 4-10 shows Enterprise Manager reporting disk failures. Three alerts appear at 11:19:29. The first alert is an Offline Disk Count. The second and third alerts are Disk Status messages for data area disk DATA.XBBT1D06_DATA and recovery area disk RECO.XBBT1D06_RECO:

2 disks are offline
Disk DATA.XBBT1D06_DATA is offline.
Disk RECO.XBBT1D06_RECO is offline.

Figure 4-10 Enterprise Manager Reports Disk Failures

Description of "Figure 4-10 Enterprise Manager Reports Disk Failures"

Figure 4-11 shows Enterprise Manager reporting the status of data area disk group DATA and recovery area disk group RECO. The red arrows under Member Disks indicate that one disk has failed in each disk group. The numbers under Pending Operations indicate that one operation is pending for each disk group.

Figure 4-11 Enterprise Manager Reports ASM Disk Groups Status

Description of "Figure 4-11 Enterprise Manager Reports ASM Disk Groups Status"

Figure 4-12 shows Enterprise Manager reporting a pending REBAL operation on the DATA disk group. The operation is about one-third done, as shown in % Complete, and the Remaining Time is estimated to be 16 minutes.

Figure 4-12 Enterprise Manager Reports Pending REBAL Operation

Description of "Figure 4-12 Enterprise Manager Reports Pending REBAL Operation"

4.2.6.3 Data Area Disk Group Failure

A data area disk group failure should occur only when there have been multiple failures. For example, if the data-area disk group is defined as external redundancy, a single-disk failure should not be exposed to ASM. However, multiple disk failures in a storage array may be seen by ASM causing the disk group to go offline. Similarly, multiple disk failures in different failure groups in a normal or high-redundancy disk group may cause the disk group to go offline.

When one or more disks fail in a normal or high redundancy disk group, but the ASM disk group is accessible, there is no loss of data and no immediate loss of accessibility. An ASM instance automatically starts an ASM rebalance operation to distribute the data on the one or more failed disks to alternative disks in the ASM disk group. When the rebalance operation completes successfully, the ASM disk group is no longer at risk in the event of a second failure. There must be enough disk space on the remaining disks in the disk group for the rebalance to complete successfully.

The two possible solutions when the data area disk group fails are summarized in Table 4-8.

If Data Guard is being used and fast-start failover is configured, then an automatic failover will occur when the database shuts down due to the data-area disk group going offline. If fast-start failover is not configured, then perform a manual failover.

If you decide to perform a Data Guard failover, then the recovery time objective (RTO) will be expressed in terms of minutes or perhaps seconds, depending on the presence of the Data Guard observer process and fast-start failover. However, if a manual failover occurs and not all data is on the standby site, then data loss might result.

After Data Guard failover has completed, and the application is available, the data area disk group failure must still be resolved. Continue with the following "Local Recovery Steps" procedure to resolve the ASM disk group failure.

The RTO for local recovery only is based primarily on the time required to:

1. Repair and replace the failed storage components

2. Restore and recover the database

Because the loss affects only the data-area disk group, there is no loss of data. All transactions are recorded in the Oracle redo log members that reside in the flash recovery area, so complete media recovery is possible.

If Data Guard is not being used, then perform the following local recovery steps. The time required to perform local recovery depends on how long it takes to restore and recover the database. There is no data loss when performing local recovery.

Local Recovery Steps

Perform these steps after one or more failed disks have been replaced and access to the storage has been restored:

1. Rebuild the ASM disk group using the new storage location:

   SQL> CREATE DISKGROUP DATA NORMAL REDUNDANCY DISK 'path1','path2',...;
   
   

2. Start the instance NOMOUNT:

   RMAN> STARTUP FORCE NOMOUNT;
   
   

3. Restore the control file from the surviving copy located in the recovery area:

   RMAN> RESTORE CONTROLFILE FROM 'recovery_area_controlfile';
   
   

4. Start the instance MOUNT:

   RMAN> STARTUP FORCE MOUNT;
   
   

5. Restore the database:

   RMAN> RESTORE DATABASE
   
   

6. Recover the database:

   RMAN> RECOVER DATABASE;
   
   

7. If you use block change tracking, then disable and reenable the block change tracking file:

   SQL> ALTER DATABASE DISABLE BLOCK CHANGE TRACKING;
   SQL> ALTER DATABASE ENABLE BLOCK CHANGE TRACKING;
   
   

8. Open the database:

   SQL> ALTER DATABASE OPEN;
   
   

9. Re-create the log file members on the failed ASM disk group:

   SQL> ALTER DATABASE DROP LOGFILE MEMBER 'filename';
   SQL> ALTER DATABASE ADD LOGFILE MEMBER 'disk_group' TO GROUP group_no;
   
   

10. Perform a new incremental level 0 backup:

    RMAN> BACKUP INCREMENTAL LEVEL 0 DATABASE;
    

4.2.6.4 Flash Recovery Area Disk Group Failure

When the flash recovery-area disk group fails, the database crashes because the control file member usually resides in the flash recovery area and Oracle requires that all control file members are accessible. The flash recovery area can also contain the flashback logs, redo log members and all backups.

A flash recovery area disk group failure should occur only when there have been multiple failures. For example, if the flash recovery-area disk group is defined as external redundancy, a single-disk failure should not be exposed to ASM. However, multiple-disk failures in a storage array may be seen by ASM causing the disk group to go offline. Similarly, multiple-disk failures in different failure groups in a normal or high-redundancy disk group may cause the disk group to go offline.

Table 4-9 summarizes the two possible solutions when the flash recovery-area disk group fails.

Because the loss affects only the flash recovery-area disk group, there is no loss of data. No database media recovery is required, because the data files and the online redo log files are still present and available in the data area. A fast local restart is to startup the primary database after removing the controlfile member located in the failed flash recovery area and point to a new flash recovery area for local archiving (see "Local Restart Steps" discussion later in this section for the step-by-step procedure). However, this is a temporary fix until a new flash recovery area is created to replace the failed storage components. Oracle recommends using the "Local Recovery Steps" discussion later in this section.

If you decide to perform a Data Guard failover, then the RTO will be expressed in terms of minutes or perhaps seconds depending on the presence of the Data Guard observer process and fast-start failover. After Data Guard failover has completed, and the application is available, the flash recovery area disk group failure must still be resolved. Continue with the instructions in the following "Local Recovery Steps" section to resolve the ASM disk group failure.

If the protection level is maximum performance or the standby database is unsynchronized with the primary database, then temporarily start up the primary database by removing the controlfile member and pointing to a temporary flash recovery area (file system) in the SPFILE. Issue a Data Guard switchover to ensure no data loss. After Data Guard switchover has completed, and the application is available, the flash recovery area disk group failure must still be resolved. Shut down the affected database and continue with the instructions in the following "Local Recovery Steps" section to resolve the ASM disk group failure.

The RTO for local recovery only is based primarily on the time to repair and replace the failed storage components and then on the time to restore the control-file copy. Because the loss affects only the flash recovery-area disk group, there is no loss of data. No database media recovery is required, because the data files and the online redo log files are still present and available in the data area. As mentioned previously, you can start up the primary database by removing the controlfile member and pointing to a new flash recovery area. However, this is a temporary fix filled with availability and performance risks unless the flash recovery area is configured properly. Therefore, Oracle recommends the "Local Recovery Steps" that follow.

Local Restart Steps

For a fast local restart, perform the following steps on the primary database:

1. Change the CONTROL_FILES initialization parameter to refer only to members in the Data Area. For example:

   ALTER SYSTEM SET CONTROL_FILES='+DATA/sales/control1.dbf' SCOPE=spfile;
   
   

2. Change local archive destinations and/or the flash recovery area to the local redundant, scalable destination. For example:

   ALTER SYSTEM SET DB_RECOVERY_FILE_DEST='+DATA' SCOPE=spfile;
   
   

3. Startup with new settings:

   STARTUP MOUNT:
   
   

You may need to disable and reenable Flashback Database because the flashback logs were damaged or lost:

ALTER DATABASE FLASHBACK OFF;
ALTER DATABASE FLASHBACK ON;
ALTER DATABASE OPEN;

Local Recovery Steps

1. Replace or get access to new storage to be leveraged as flash recovery area

2. Rebuild the ASM disk group using the new storage location:

   SQL> CREATE DISKGROUP RECO NORMAL REDUNDANCY DISK 'path1','path2',...;
   
   

3. Start the instance NOMOUNT:

   RMAN> STARTUP FORCE NOMOUNT;
   
   

4. Restore the control file from the surviving copy located in the data area:

   RMAN> RESTORE CONTROLFILE FROM 'data_area_controlfile';
   
   

5. Start the instance MOUNT:

   RMAN> STARTUP FORCE MOUNT;
   
   

6. If you use Flashback Database, then disable it:

   SQL> ALTER DATABASE FLASHBACK OFF;
   
   

7. Open the database and allow instance recovery to complete:

   SQL> ALTER DATABASE OPEN;
   
   

8. Issue the following statements only if Flashback Database is required:

   SQL> SHUTDOWN IMMEDIATE;
   SQL> STARTUP MOUNT;
   SQL> ALTER DATABASE FLASHBACK ON;
   SQL> ALTER DATABASE OPEN;
   
   

9. Re-create the log file members on the failed ASM disk group:

   SQL> ALTER DATABASE DROP LOGFILE MEMBER 'filename';
   SQL> ALTER DATABASE ADD LOGFILE MEMBER 'disk_group' TO GROUP group_no;
   
   

10. Synchronize the control file and the flash recovery area:

    RMAN> CATALOG RECOVERY AREA;
    RMAN> CROSSCHECK ARCHIVELOG ALL;
    RMAN> CROSSCHECK BACKUPSET;
    RMAN> CROSSCHECK DATAFILECOPY ALL;
    RMAN> LIST EXPIRED type;
    RMAN> DELETE EXPIRED type;
    
    

11. Assuming that data has been lost in some way, take a new backup:

    RMAN> BACKUP INCREMENTAL LEVEL 0 DATABASE;
    

4.2.7.1 Use Data Guard to Recover From Data Corruption and Data Failure

Failover is the operation of transitioning the standby databases as the new production database. A database switchover is a planned transition in which a standby database and a production database switch roles. Either of these operations can occur in less than 5 minutes and with no data loss.Use Data Guard switchover or failover for data corruption or data failure when:

• The database is down or when the database is up but the application is unavailable because of data corruption or failure, and the time to restore and recover locally is long or unknown.

• Recovering locally will be longer than the business SLA or RTO.

4.2.7.2 Use RMAN Block Media Recovery

Block media recovery (BMR) recovers one block or a set of data blocks marked "media corrupt" in a data file by using the RMAN BLOCKRECOVER command. When a small number of data blocks are marked media corrupt and require media recovery, you can selectively restore and recover damaged blocks rather than whole data files. This results in lower RTO because only blocks that need recovery are restored and only necessary corrupt blocks undergo recovery. Block media recovery minimizes redo application time and avoids I/O overhead during recovery. It also enables affected data files to remain online during recovery of the corrupt blocks. The corrupt blocks, however, remain unavailable until they are completely recovered.

Use block media recovery when:

• A small number of blocks require media recovery and the blocks that need recovery are known. If a significant portion of the datafile is corrupt, or if the amount of corruption is unknown, then a different recovery method should be used.

• Blocks are marked corrupt (verified with the RMAN BACKUP VALIDATE command) and only when complete recovery is required.

• Backup of the data file containing the corrupted blocks is available locally or can be retrieved from a remote location including from a a physical standby database.

Block media recovery cannot be used to recover from the following:

• User error or software bugs that cause logical corruption where the data blocks are intact. See Section 4.2.8, "Recovering from Human Error" for additional details for this type of recovery.

• Changes caused by corrupt redo data. Block media recovery requires that all available redo data be applied to the blocks being recovered.

For example, to recover a specific corrupt block using RMAN block media recovery:

RMAN> BLOCKRECOVER DATAFILE 7 BLOCK 3;

When the corruption is detected, it would be easy to recover this block through Grid Control.

4.2.7.3 Use RMAN Data File Media Recovery

Data file media recovery recovers an entire datafile or set of data files for a database by using the RMAN RECOVER command. When a large or unknown number of data blocks are marked media-corrupt and require media recovery, or when an entire file is lost, the affected data files must be restored and recovered.

Use RMAN file media recovery when the following conditions are true:

• The number of blocks requiring recovery is large or unknown

• Block media recovery is not available (for example, if incomplete recovery is required, or if only incremental backups are available for the data file requiring recovery)

  See Also:

  "Advanced User-Managed Recovery Scenarios" in Oracle Database Backup and Recovery Advanced User's Guide and the white paper titled Using Recovery Manager with Oracle Data Guard in Oracle Database 10g available at http://www.oracle.com/technology/deploy/availability/pdf/RMAN_DataGuard_10g_wp.pdf

4.2.7.4 Re-Create Objects Manually

Some database objects, such as small look-up tables or indexes, can be recovered quickly by manually re-creating the object instead of doing media recovery.

Use manual object re-creation when:

• You must re-create a small index because of media corruption. Creating an index online enables the base object to be used concurrently.

• You must re-create a look-up table or when the scripts to re-create the table are readily available. Dropping and re-creating the table might be the fastest option.

4.2.8.1 Resolving Table Inconsistencies

Oracle provides a FLASHBACK DROP statement to recover from an accidental DROP TABLE statement, and a FLASHBACK TABLE statement to restore a table to a previous point in the database.

Flashback Table provides the DBA the ability to recover a table, or a set of tables, to a specified point in time quickly and easily. In many cases, Flashback Table alleviates the more complicated point in time recovery operations. For example:

FLASHBACK TABLE orders, order_items 
      TO TIMESTAMP 
      TO_DATE('28-Jun-06 14.00.00','dd-Mon-yy hh24:mi:ss');

This statement rewinds any updates to the ORDERS and ORDER_ITEMS tables that have been done between the current time and specified timestamp in the past. Flashback Table performs this operation online and in place, and it maintains referential integrity constraints between the tables.

Dropping or deleting database objects by accident is a common mistake. Users soon realize their mistake, but by then it is too late and there has been no way to easily recover the dropped tables and its indexes, constraints, and triggers. Objects once dropped were dropped forever. Loss of very important tables or other objects (like indexes, partitions or clusters) required DBAs to perform a point-in-time recovery, which can be time-consuming and lead to loss of recent transactions.

Flashback Drop provides a safety net when dropping objects in Oracle Database 10g. When a user drops a table, Oracle places it in a recycle bin. Objects in the recycle bin remain there until the user decides to permanently remove them or until space limitations begin to occur on the tablespace containing the table. The recycle bin is a virtual container where all dropped objects reside. Users view the recycle bin and undrop the dropped table and its dependent objects. For example, the employees table and all its dependent objects would be undropped by the following statement:

FLASHBACK TABLE employees TO BEFORE DROP;

4.2.8.2 Resolving Row and Transaction Inconsistencies

Resolving row and transaction inconsistencies might require a combination of Flashback Query, Flashback Version Query, Flashback Transaction Query, and the compensating SQL statements constructed from undo statements to rectify the problem. This section describes a general approach using a human resources example to resolve row and transaction inconsistencies caused by erroneous or malicious user errors.

Flashback Query, a feature introduced in the Oracle9i Database, enables an administrator or user to query any data at some point in time in the past. This powerful feature can be used to view and reconstruct data that might have been deleted or changed by accident. For example:

SELECT * FROM EMPLOYEES 
       AS OF TIMESTAMP 
       TO_DATE('28-Jun-06 14:00','DD-Mon-YY HH24:MI')
WHERE ...

This partial statement displays rows from the EMPLOYEES table starting from 2 p.m. on June 28, 2006. Developers can use this feature to build self-service error correction into their applications, empowering end users to undo and correct their errors without delay, rather than burdening administrators to perform this task. Flashback Query is very simple to manage, because the database automatically keeps the necessary information to reconstruct data for a configurable time into the past.

Flashback Version Query provides a way to view changes made to the database at the row level. It is an extension to SQL and enables the retrieval of all the different versions of a row across a specified time interval. For example:

SELECT * FROM EMPLOYEES
       VERSIONS BETWEEN TIMESTAMP
       TO_DATE('28-Jun-06 14:00','dd-Mon-YY hh24:mi') AND
       TO_DATE('28-Jun-06 15:00','dd-Mon-YY hh24:mi')
WHERE ...

This statement displays each version of the row, each entry changed by a different transaction, between 2 and 3 p.m. today. A DBA can use this to pinpoint when and how data is changed and trace it back to the user, application, or transaction. This enables the DBA to track down the source of a logical corruption in the database and correct it. It also enables application developers to debug their code.

Flashback Transaction Query provides a way to view changes made to the database at the transaction level. It is an extension to SQL that enables you to see all changes made by a transaction. For example:

SELECT UNDO_SQL
FROM FLASHBACK_TRANSACTION_QUERY 
WHERE XID = '000200030000002D';

This statement shows all of the changes that resulted from this transaction. In addition, compensating SQL statements are returned and can be used to undo changes made to all rows by this transaction. Using a precision tool like this, the DBA and application developer can precisely diagnose and correct logical problems in the database or application.

Consider a human resources (HR) example involving the SCOTT schema. The HR manager reports to the DBA that there is a potential discrepancy in Ward's salary. Sometime before 9:00 a.m., Ward's salary was increased to $1875. The HR manager is uncertain how this occurred and wishes to know when the employee's salary was increased. In addition, he instructed his staff to reset the salary to the previous level of $1250. This was completed around 9:15 a.m.

The following steps show how to approach the problem.

1. Assess the problem.

   Fortunately, the HR manager has provided information about the time when the change occurred. You can query the information as it was at 9:00 a.m. with Flashback Query.

   SELECT EMPNO, ENAME, SAL
       FROM EMP
       AS OF TIMESTAMP TO_DATE('28-JUN-06 09:00','dd-Mon-yy hh24:mi')
       WHERE ENAME = 'WARD';
   
            EMPNO ENAME             SAL
       ---------- ---------- ----------
             7521 WARD             1875
   
   

   To can confirm that you have the correct employee by the fact that Ward's salary was $1875 at 09:00 a.m. Rather than using Ward's name, you can now use the employee number for subsequent investigation.

2. Query past rows or versions of the data to acquire transaction information.

   Although it is possible to restrict the row version information to a specific date or SCN range, you might want to query all the row information that is available for the employee WARD using Flashback Version Query.

   SELECT EMPNO, ENAME, SAL, VERSIONS_STARTTIME, VERSIONS_ENDTIME
       FROM EMP
       VERSIONS BETWEEN TIMESTAMP MINVALUE AND MAXVALUE
       WHERE EMPNO = 7521
       ORDER BY NVL(VERSIONS_STARTSCN,1);
   
   EMPNO ENAME          SAL VERSIONS_STARTTIME     VERSIONS_ENDTIME
   ----- ---------- ------- ---------------------- ----------------------
    7521 WARD          1250  28-JUN-06 08.48.43 AM  28-JUN-06 08.54.49 AM
    7521 WARD          1875  28-JUN-06 08.54.49 AM  28-JUN-06 09.10.09 AM
    7521 WARD          1250  28-JUN-06 09.10.09 AM
   
   

   You can see that WARD's salary was increased from $1250 to $1875 at 08:54:49 the same morning and was subsequently reset to $1250 at approximately 09:10:09.

   Also, you can modify the query to determine the transaction information for each of the changes affecting WARD using a similar Flashback Version Query. This time use the VERSIONS_XID pseudocolumn.

   SELECT EMPNO, ENAME, SAL, VERSIONS_XID
       FROM EMP
       VERSIONS BETWEEN TIMESTAMP MINVALUE AND MAXVALUE
       WHERE EMPNO = 7521
       ORDER BY NVL(VERSIONS_STARTSCN,1);
   
            EMPNO ENAME             SAL VERSIONS_XID
       ---------- ---------- ---------- ----------------
             7521 WARD             1250 0006000800000086
             7521 WARD             1875 0009000500000089
             7521 WARD             1250 000800050000008B
   
   

   The ID of the erroneous transaction that increased WARD's salary to $1875 was "0009000500000089".

3. Query the erroneous transaction and the scope of its effect.

   With the transaction information (VERSIONS_XID pseudocolumn), you can now query the database to determine the scope of the transaction, using Flashback Transaction Query.

   SELECT UNDO_SQL
       FROM FLASHBACK_TRANSACTION_QUERY
       WHERE XID = HEXTORAW('0009000500000089');
   
       UNDO_SQL                                                                    
       ----------------------------------------------------------------------------
       update "SCOTT"."EMP" set "SAL" = '950' where ROWID = 'AAACV4AAFAAAAKtAAL';      
       update "SCOTT"."EMP" set "SAL" = '1500' where ROWID = 'AAACV4AAFAAAAKtAAJ';     
       update "SCOTT"."EMP" set "SAL" = '2850' where ROWID = 'AAACV4AAFAAAAKtAAF';      
       update "SCOTT"."EMP" set "SAL" = '1250' where ROWID = 'AAACV4AAFAAAAKtAAE';    
       update "SCOTT"."EMP" set "SAL" = '1600' where ROWID = 'AAACV4AAFAAAAKtAAB';     
                                                                                   
       6 rows selected.
   
   

   You can see that WARD's salary was not the only change that occurred in the transaction. The information that was changed for the other four employees at the same time as WARD can now be passed back to the HR manager for review.

4. Determine if the corrective statements should be executed.

   If the HR manager decides that the corrective changes suggested by the UNDO_SQL column are correct, then the DBA can execute those statements individually.

5. Query the FLASHBACK_TRANSACTION_QUERY view for additional transaction information. For example, to determine the user that performed the erroneous update, issue the following query:

   SELECT LOGON_USER FROM FLASHBACK_TRANSACTION_QUERY
   WHERE XID = HEXTORAW('0009000500000089');
   
   LOGON_USER
   ----------------------------
   MSMITH
   
   

   In this example, the query shows that the user MSMITH was responsible for the erroneous transaction.

4.2.8.3 Resolving Database-Wide Inconsistencies

To bring an Oracle database to a previous point in time, the traditional method is point-in-time recovery. However, point-in-time recovery can take hours or even days, because it requires the whole database to be restored from backup and recovered to the point in time just before the error was introduced into the database. With the size of databases constantly growing, it will take hours or even days just to restore the whole database.

Flashback Database is a new strategy for doing point-in-time recovery. It quickly rewinds an Oracle database to a previous time to correct any problems caused by logical data corruption or user error. Flashback logs are used to capture old versions of changed blocks. One way to think of it is as a continuous backup or storage snapshot. When recovery must be performed the flashback logs are quickly replayed to restore the database to a point in time before the error and just the changed blocks are restored. It is extremely fast and reduces recovery time from hours to minutes. In addition, it is easy to use. A database can be recovered to 2:05 p.m. by issuing a single statement. Before the database can be recovered, all instances of the database must be shut down and one of the instances subsequently mounted. The following is an example of a FLASHBACK DATABASE statement.

FLASHBACK DATABASE TO TIMESTAMP SYSDATE-1;

No restoration from tape, no lengthy downtime, and no complicated recovery procedures are required to use it. You can also use Flashback Database and then open the database in read-only mode and examine its contents. If you determine that you flashed back too far or not far enough, then you can reissue the FLASHBACK DATABASE statement or continue recovery to a later time to find the proper point in time before the database was damaged. Flashback Database works with a production database, a physical standby database, or a logical standby database.

These steps are recommended for using Flashback Database:

1. Determine the time or the SCN to which to flash back the database.

2. Verify that there is sufficient flashback log information.

   SELECT OLDEST_FLASHBACK_SCN, 
          TO_CHAR(OLDEST_FLASHBACK_TIME, 'mon-dd-yyyy HH:MI:SS') 
          FROM V$FLASHBACK_DATABASE_LOG;
   
   

3. Flash back the database to a specific time or SCN. (The database must be mounted to perform a Flashback Database.)

   FLASHBACK DATABASE TO SCN scn;
   
   

   or

   FLASHBACK DATABASE TO TIMESTAMP TO_DATE date;
   
   

4. Open the database in read-only mode to verify that it is in the correct state.

   ALTER DATABASE OPEN READ ONLY;
   
   

   If more flashback data is required, then issue another FLASHBACK DATABASE statement. (The database must be mounted to perform a Flashback Database.)

   If you want to move forward in time, then issue a statement similar to the following:

   RECOVER DATABASE UNTIL [TIME | CHANGE] date | scn;
   
   

5. Open the database:

   ALTER DATABASE OPEN RESETLOGS;
   
   

Other considerations when using Flashback Database are as follows:

• If there are not sufficient flashback logs to flash back to the target time, then use one of the following alternatives:

  • Use Data Guard to recover to the target time if the standby lags behind the primary database or flash back to the target time if there's sufficient flashback logs on the standby.

  • Restore from backups.

• After flashing back a database, any dependent database such as a standby database must be flashed back. See Section 4.3, "Restoring Fault Tolerance".

Flashback Database does not automatically fix a dropped tablespace, but it can be used to dramatically reduce the downtime. You can flash back the production database to a point before the tablespace was dropped and then restore a backup of the corresponding datafiles from the affected tablespace and recover to a time before the tablespace was dropped.

Follow these recommended steps to use Flashback Database to repair a dropped tablespace:

1. Determine the SCN or time you dropped the tablespace.

2. Flash back the database to a time before the tablespace was dropped. You can use a statement similar to the following:

   FLASHBACK DATABASE TO BEFORE SCN drop_scn;
   
   

3. Restore, rename, and bring datafiles online.

   1. Restore only the datafiles from the affected tablespace from a backup.

   2. Rename the unnamed files to the backup files.

   ALTER DATABASE RENAME FILE '.../UNNAMED00005' to 'restored_file';
   
   

4. Bring the datafiles online.

   ALTER DATABASE DATAFILE 'name' ONLINE; 
            
   

5. Query and recover the database.

   SELECT CHECKPOINT_CHANGE# FROM V$DATAFILE_HEADER WHERE FILE#=1;
   RECOVER DATABASE UNTIL CHANGE checkpoint_change#;
   
   

6. Open the database.

   ALTER DATABASE OPEN RESETLOGS;
   

4.3.1.1 Recovering Service Availability

After a failed node has been brought back into the cluster and its instance has been started, Cluster Ready Services (CRS) automatically manages the virtual IP address used for the node and the services supported by that instance automatically. A particular service might or might not be started for the restored instance. The decision by CRS to start a service on the restored instance depends on how the service is configured and whether the proper number of instances are currently providing access for the service. A service is not relocated back to a preferred instance if the service is still being provided by an available instance to which it was moved by CRS when the initial failure occurred. CRS restarts services on the restored instance if the number of instances that are providing access to a service across the cluster is less than the number of preferred instances defined for the service. After CRS restarts a service on a restored instance, CRS notifies registered applications of the service change.

For example, suppose the HR service is defined with instances A and B as preferred and instances C and D as available in case of a failure. If instance B fails and CRS starts up the HR service on C automatically, then when instance B is restarted, the HR service remains at instance C. CRS does not automatically relocate a service back to a preferred instance.

Suppose a different scenario in which the HR service is defined with instances A, B, C, and D as preferred and no instances defined as available, spreading the service across all nodes in the cluster. If instance B fails, then the HR service remains available on the remaining three nodes. CRS automatically starts the HR service on instance B when it rejoins the cluster because it is running on fewer instances than configured. CRS notifies the applications that the HR service is again available on instance B.

4.3.1.2 Considerations for Client Connections After Restoring a RAC Instance

After a RAC instance has been restored, additional steps might be required, depending on the current resource utilization and performance of the system, the application configuration, and the network load balancing that has been implemented.

Existing connections (which might have failed over or started as a new session) on the surviving RAC instances, are not automatically redistributed or failed back to an instance that has been restarted. Failing back or redistributing users might or might not be necessary, depending on the current resource utilization and the capability of the surviving instances to adequately handle and provide acceptable response times for the workload. If the surviving RAC instances do not have adequate resources to run a full workload or to provide acceptable response times, then it might be necessary to move (disconnect and reconnect) some existing user connections to the restarted instance.

New connections are started as they are needed, on the least-used node, assuming connection load balancing has been configured. Therefore, the new connections are automatically load-balanced over time.

An application service can be:

• Partitioned with services running on a subset of RAC instances

• Nonpartitioned so that all services run equally across all nodes

This is valuable for modularizing application and database form and function while still maintaining a consolidated data set. For the cases where an application is partitioned or has a combination of partitioning and nonpartitioning, the response time and availability aspects for each service should be considered. If redistribution or failback of connections for a particular service is required, then you can rebalance workloads manually with the DBMS_SERVICE.DISCONNECT_SESSION PL/SQL procedure. You can use this procedure to disconnect sessions associated with a service while the service is running.

For load-balancing application services across multiple RAC instances, Oracle Net connect-time failover and connection load balancing are recommended. This feature does not require changes or modifications for failover or restoration. It is also possible to use hardware-based load balancers. However, there might be limitations in distinguishing separate application services (which is understood by Oracle Net Services) and restoring an instance or a node. For example, when a node or instance is restored and available to start receiving new connections, a manual step might be required to include the restored node or instance in the hardware-based load balancer logic, whereas Oracle Net Services does not require manual reconfiguration.

Table 4-14 summarizes the considerations for new and existing connections after an instance has been restored. The considerations differ depending on whether the application services are partitioned, nonpartitioned, or are a combination of both. The actual redistribution of existing connections might or might not be required depending on the resource utilization and response times.

Figure 4-13 shows a two-node partitioned RAC database. Each instance services a different portion of the application (HR and Sales). Client processes connect to the appropriate instance based on the service they require.

Figure 4-13 Partitioned Two-Node RAC Database

Description of "Figure 4-13 Partitioned Two-Node RAC Database"

Figure 4-14 shows what happens when one RAC instance fails.

Figure 4-14 RAC Instance Failover in a Partitioned Database

Description of "Figure 4-14 RAC Instance Failover in a Partitioned Database"

If one RAC instance fails, then the service and existing client connections can be automatically failed over to another RAC instance. In this example, the HR and Sales services are both supported by the remaining RAC instance. In addition, new client connections for the Sales service can be routed to the instance now supporting this service.

After the failed instance has been repaired and restored to the state shown in Figure 4-13 and the Sales service is relocated to the restored instance failed-over clients and any new clients that had connected to the Sales service on the failed-over instance might have to be identified and failed back. New client connections, which are started after the instance has been restored, should automatically connect back to the original instance. Therefore, over time, as older connections disconnect, and new sessions connect to the Sales service, the client load migrates back to the restored instance. Rebalancing the load immediately after restoration depends on the resource utilization and application response times.

Figure 4-15 shows a nonpartitioned application. Services are evenly distributed across both active instances. Each instance has a mix of client connections for both HR and Sales.

Figure 4-15 Nonpartitioned RAC Instances

Description of "Figure 4-15 Nonpartitioned RAC Instances"

If one RAC instance fails, then CRS moves the services that were running on the failed instance. In addition, new client connections are routed only to the available RAC instances that offer that service.

After the failed instance has been repaired and restored to the state shown in Figure 4-15, some clients might have to be moved back to the restored instance. For nonpartitioned applications, identifying appropriate services is not required for rebalancing the client load among all available instances. Also, this is necessary only if a single instance is not able to adequately service the requests.

New client connections that are started after the instance has been restored should automatically connect back to the restored instance because it has a smaller load. Therefore, over time, as older connections disconnect and new sessions connect to the restored instance, the client load will again evenly balance across all available RAC instances. Rebalancing the load immediately after restoration depends on the resource utilization and application response times.

4.3.2.1 Restoring a Standby Database After a Fast-Start Failover

Following a fast-start failover, the observer periodically attempts to reconnect to the original primary database. When the observer regains network access to the original primary database, it initiates a request for the Data Guard broker to automatically reinstate it as a standby database to the new primary. This quickly restores disaster protection and high availability for the new primary database.

You can enable fast-start failover from any site, including the observer site, in Enterprise Manager while connected to any database in the broker configuration. The broker simplifies switchovers and failovers by allowing you to invoke them using a single key click in Oracle Enterprise Manager, as shown in Figure 4-16.

Figure 4-16 Fast-Start Failover and the Observer Are Successfully Enabled

Description of "Figure 4-16 Fast-Start Failover and the Observer Are Successfully Enabled"

4.3.2.2 Reinstating a Standby Database Using Enterprise Manager After a Failover

Furthermore, you can leverage Enterprise Manager to reinstate the old primary as the new standby. Figure 4-17 shows an example of the warning message that shows in Enterprise Manager when a reinstatement is needed.

Figure 4-17 Reinstating the Former Primary Database After a Fast-Start Failover

Description of "Figure 4-17 Reinstating the Former Primary Database After a Fast-Start Failover"

4.4.1.1 Migrating to ASM Storage

If you have an existing Oracle database that stores database files on a file system or on raw devices, you can migrate some or all of these database files to ASM. The DBMS_FILE_TRANSFER package can be used for this purpose.

4.4.1.2 Adding and Removing Storage

Disks can be added to and removed from ASM with no downtime. When disks are added or removed, ASM automatically starts a rebalance operation to evenly spread the disk group contents over all drives in the disk group.The best practices for adding or removing storage include:

• Investigate methods of adding storage to, and removing storage from, the host operating system with no downtime.

• Use a single ALTER DISKGROUP command when adding or removing multiple disk drives.

  For example, if the storage maintenance is to add new drives and remove existing drives, use a single ALTER DISKGROUP command with the ADD DISK clause to add the new drives, and the DROP DISK clause to remove the existing drives. For example:

  ALTER DISKGROUP data
         DROP DISK diska5
         ADD FAILGROUP failgrp1 DISK '/devices/diska9' NAME diska9;
  
  

• When dropping disks from a disk group, specify the WAIT option in the REBALANCE clause so the ALTER DISKGROUP statement does not return until the contents of the drives being dropped have been moved to other drives. Once the statement completes, the drives can be safely removed from the system. For example:

  ALTER DISKGROUP data
  DROP DISK diska5
  ADD FAILGROUP failgrp1 DISK '/devices/diska9' NAME diska9
  REBALANCE WAIT;
  
  

• When dropping disks in a normal or high redundancy disk group, ensure there is enough free disk space in the disk group to reconstruct full redundancy.

• Monitor the progress of rebalance operations using Enterprise Manager or by querying V$ASM_OPERATION.

• For long-running rebalance operations that occur during periods of low database activity, increase the rebalance power limit to reduce the rebalance time.

4.4.2.1 Best Practices To Minimize Downtime

Use the following recommended practices for all database patch upgrades:

• Always confirm with Oracle Support Services that the patch is valid for your problem and for your deployment environment.

• Have a plan for applying the patch as well as a plan for backing out the patch.

• Apply the patch to your test environment first and verify that it fixes the problem.

• When you plan the elapsed time for applying the patch, include time for starting up and shutting down the other tiers of your technology stack if necessary.

• If the patch is not a candidate for RAC rolling upgrade and you can incur the downtime for applying the patch, go to Section 4.4.3, "Database Upgrades" to assess whether or not other solutions are feasible.

The following are additional recommended practices for RAC rolling upgrades.

• If multiple instances share an Oracle home, then all of them will be affected by application of a patch. The DBA should verify that this will not cause unintentional side effects. Also, all such instances on a node must be shut down during the patch application. Scheduled outage planning should take this into account. As a best practice, only similar applications should share an Oracle home on a node. This provides greater flexibility for patching.

• The Oracle inventory on each node is a repository of the Oracle Database software installed on the node. The inventory is node-specific. It is shared by all Oracle software installed on the node. It is similar across nodes only if all nodes are exactly the same in terms of the Oracle Database software deployed, the deployment configuration, and patch levels. Because the Oracle inventory greatly aids the patch application and patch management process, it is recommended that its integrity be maintained. Oracle inventory should be backed up after each patch installation to any Oracle software on a specific node. This applies to the Oracle inventory on each node of the cluster.

• Use the Oracle Universal Installer to install all Oracle database software. This creates the relevant repository entries in the Oracle inventory on each node of the cluster. Also, use the Oracle Universal Installer to add nodes to an existing RAC cluster.

  However, if this was not done or is not feasible for some reason, adding information about an existing Oracle database software installation to the Oracle inventory can be done with the attach option of the opatch utility. Node information can be also added with this option.

• The nature of the rolling patch upgrade enables it to be applied to only some nodes of the RAC cluster. So an instance can be operating with the patch applied, while another instance is operating without the patch. This is not possible for nonrolling patch upgrades. Apply nonrolling patch upgrades to all instances before the RAC deployment is activated. A mixed environment is useful if a patch must be tested before deploying it to all the instances. Applying the patch with the -local option is the recommended way to do this.

  In the interest of keeping all instances of the RAC cluster at the same patch level, it is strongly recommended that after a patch has been validated, it should be applied to all nodes of the RAC installation. When instances of a RAC cluster have similar patch software, services can be migrated among instances without running into the problem a patch might have fixed.

• All patches (including those applied by rolling upgrades) should be maintained online and not removed once they have been applied. This is useful if a patch must be rolled back or applied again.

  The patches should be stored in a location that is accessible by all nodes of the cluster. Thus all nodes of the cluster are equivalent in their capability to apply or roll back a patch.

• Rolling patch upgrades, just like any other patch upgrade, should be done when no other patch upgrade or Oracle installation is being done on the node. Application of multiple patches is a sequential process. The scheduled outage should be planned accordingly.

• If multiple patches have to be applied and they must be applied at the same time, and if only some of these patches are eligible for rolling upgrade, then apply all of them in a nonrolling manner. This reduces the overall time required to get through the patching process.

• For patches that are not eligible for rolling upgrade, the next best option for RAC deployments is the minimize_downtime option of the apply command.

• Perform the rolling upgrade when system usage is low. This ensures minimal disruption of service for the end user.

4.4.3.1 Database Upgrade Assistant

Database Upgrade Assistant (DBUA) is used to upgrade a database in place from an earlier software version.

When deciding if DBUA is the proper tool to use when performing a database upgrade with minimal downtime, consider the following:

• DBUA upgrades the database dictionary and all components (for example: Java, XDB, Streams, and so on) that have been installed while the database is unavailable for normal user activity.

• Downtime required for a database upgrade when using DBUA is determined by the time needed to:

  • Upgrade all database dictionary objects to the new version

  • Recompile all PL/SQL

  • Reconnect the clients to the upgraded database

Use DBUA for a database upgrade when the time to perform the upgrade with this method fits within the maintenance window.

4.4.3.2 Data Guard SQL Apply (Logical Standby)

Data Guard SQL Apply can be used to upgrade a database with minimal downtime by means of a process called rolling upgrade. Data Guard currently supports homogeneous environments where the primary and standby databases are running on the same platform.

Note the following points when deciding if Data Guard SQL Apply is the appropriate method for minimizing downtime during a database upgrade:

• The Data Guard SQL Apply infrastructure uses Oracle Streams and therefore inherits Oracle Streams data type restrictions on user-defined types, such as object types, REF values, varrays, and nested tables.

• Support for rolling upgrade starts with Oracle Database 10g release 1 (10.1.0.3). The supported versions, for both the source database and the target database, are more restrictive than Oracle Streams.

• Downtime required for a database upgrade (rolling upgrade) when using Data Guard SQL Apply is determined by the time needed to:

  • Perform a Data Guard switchover

  • Reconnect the clients to the new database

Use Data Guard SQL Apply for rolling database upgrade when DBUA will not complete the upgrade within the maintenance window and the application does not use user-defined types.

4.4.3.3 Oracle Streams

Oracle Streams can be used to upgrade the database software from one version to another with minimal downtime. This is because Oracle Streams supports a configuration in which the primary database and its replica are running on different database versions.

Note the following points when deciding if Oracle Streams is an appropriate method for a database upgrade:

• Oracle Streams does not support user-defined types, such as object types, REF values, varrays, and nested tables. However, shadow tables can be created on the primary database that do not have the unsupported data types and the shadow tables can be replicated.

• The source database must be running Oracle9i release 2 or higher.

• The administrative effort required to set up and maintain the Oracle Streams environment is more than if using Data Guard SQL Apply for a database upgrade.

• There might be a performance impact on the source database while the source and target databases run in parallel as changes are propagated to the target database.

• Downtime required for a database upgrade when using Oracle Streams is determined by the time needed to apply the remaining transactions in the queue and to reconnect the clients to the new database.

Consider using Oracle Streams if the application already uses Streams or when clients do not use user-defined types and the extra administrative effort is worth the opportunity for a very small outage time.

4.4.3.4 Transportable Tablespaces

Transportable tablespaces can be used to accomplish a database upgrade by transporting all user datafiles into a pre-created, prepared target database.

Note the following points when deciding if transportable tablespaces is the appropriate method for performing a database upgrade:

• The SYSTEM tablespace cannot be moved with transportable tablespaces. The target database SYSTEM tablespace contents, including user definitions and objects necessary for the application, must be built manually. Use Data Pump to move the contents of the SYSTEM tablespace.

• Downtime required for a database upgrade when using transportable tablespaces is determined by the time needed to:

  • Place the source database tablespaces in read-only mode

  • Perform a network import of the transportable metadata

  • If the target database is on a remote system, then include the time to transfer all datafiles from the source system to the target system

    The time it takes to transfer the datafiles can be reduced significantly by using a storage infrastructure that can make the datafiles available to the target system without the need to physically move the files, or by using a physical standby database.

Use transportable tablespaces to perform a database upgrade when DBUA will not complete within the maintenance window, and Oracle Streams or Data Guard SQL Apply cannot be used due to data type restrictions.

4.4.4.1 Transportable Database

Transportable Database is a new feature in Oracle Database 10g Release 2 (10.2) that is the recommended method for migrating an entire database to another platform that has the same endian format.

Note the following points when deciding if Transportable Database is the appropriate method to use when moving a database to another platform:

• Transportable Database supports moving databases between platforms that are of the same endian format

• Downtime required for a platform migration when using Transportable Database is determined by the time needed to:

  • Place the source database in read-only mode

  • Convert all data files to the new platform format

  • Transfer all data files from the source system to the target system

    You can significantly reduce this time by using a storage infrastructure that can make the data files available to the target system without the need to physically move the files.

Transportable Databases is the recommended method for migrating an entire database to another platform that has the same endian format because it's the simplest approach.

4.4.4.2 Oracle Streams

Oracle Streams can be used to move a database from one platform to another with minimal downtime. This is because Oracle Streams supports a configuration in which the primary database and its replica are running on different platforms.

Note the following points when deciding if Oracle Streams is an appropriate method for a platform migration:

• Oracle Streams does not support user-defined types, such as object types, REF values, varrays, and nested tables.

• To perform an upgrade using Oracle Streams, the source database must be running Oracle9i release 2 or higher.

• The administrative effort required to set up and maintain the Oracle Streams environment is more than if using Data Guard SQL Apply for a database upgrade.

• There might be a performance impact on the source database while the source and target databases run in parallel as changes are propagated to the target database.

• Downtime required for a platform migration when using Oracle Streams is determined by the time needed to apply the remaining transactions in the queue and to reconnect clients to the new database.

Consider using Oracle Streams when the application does not use user-defined types and the extra administrative effort is worth the opportunity for a very small outage time.

4.4.4.3 Oracle Data Pump

Oracle Data Pump technology enables very high-speed movement of data and metadata from one database to another, across different platforms and different database versions.

Note the following points when deciding if Data Pump is an appropriate method for a platform migration:

• Oracle Data Pump is available only on Oracle Database 10g Release 1 (10.1) and later releases.

• Downtime required for a platform migration when using Data Pump is determined by the time needed to perform a full database network import. A network import uses a database link between the target system and the remote source system to retrieve data and write it directly into the target system, without the use of dump files.

Use Data Pump when moving a database to a platform with different endian format when the network import time is acceptable.

4.4.4.4 Transportable Tablespaces

Transportable tablespaces can be used to accomplish a platform migration by transporting all user datafiles into a pre-created, prepared target database.

Note the following points when deciding if transportable tablespaces is the appropriate method for performing a platform migration:

• The SYSTEM tablespace cannot be moved with transportable tablespaces. the target database SYSTEM tablespace contents, including user definitions and objects necessary for the clients, must be built manually. Use Data Pump to move the necessary contents of the SYSTEM tablespace.

• Downtime required for a platform migration or database upgrade when using transportable tablespaces is determined by the time needed to:

  • Place the source database tablespaces in read-only mode

  • Perform a network import of the transportable metadata

  • Transfer all datafiles from the source system to the target system

    This time can be reduced significantly by using a storage infrastructure that can make the datafiles available to the target system without the need to physically move the files

  • Convert all datafiles to the new platform format using RMAN

Use transportable tablespaces to migrate to a new platform when Oracle Data Pump will not complete within the maintenance window, and Oracle Streams or Data Guard SQL Apply cannot be used due to data type restrictions.

4.4.4.5 Data Guard Redo Apply (Physical Standby Database)

Data Guard Redo Apply can be used to change the location of a database to a remote site with minimal downtime by setting up a temporary standby database at a remote location and performing a switchover operation.

Downtime required for a location migration when using Data Guard Redo Apply is determined by the time required to perform a switchover operation.