High availability, backup, resiliency, and recovery enhancements

Db2 12.1 includes enhancements that help ensure that your data remains available.

The following table lists high availability enhancements that are available in Db2® 12.1:
Table 1. High availability enhancements in Db2 12.1
Enhancement Description
New HADR upgrade procedure that allows read only access to HADR standby database during the upgrade procedure. Running upgrade procedures mentioned in Upgrade Db2 High Availability Disaster Recovery (HADR) environments can result in the database not being available to applications during the upgrade procedure. With Db2 12.1, a new HADR upgrade procedure provides read only access to the HADR standby database while upgrading the primary database.
For more information, see Upgrading a Db2 server in an HADR environment with down-level standby for Reads on Standby.
New registry variable for controlling how an HADR standby reacts when running into a disk full condition in the active log path. The DB2_HADR_DISKFULL_NONBLOCKING registry variable is now renamed to DB2_HADR_BLOCK_ON_DISKFULL. This registry variable controls the behavior of an HADR standby that is running into a disk full condition in the active log path.
By default, when an HADR standby runs into a disk full condition in the active log path, the standby stays up, with the STANDBY_LOG_DEVICE_FULL flag set in the HADR_FLAGS field, as reported by the db2pd -hadr HADR monitor table function. This can cause transactions on primary to be blocked.
When you set the new DB2_HADR_BLOCK_ON_DISKFULL registry variable to OFF, the standby database is brought down in a disk-full situation. Taking the standby down takes HADR out of PEER state and unblocks the transactions on the primary.
ADMIN_MOVE_TABLE provides faster move from row to columnar table format. ADMIN_MOVE_TABLE can take advantage of the parallel and vectorized insert when moving row tables to columnar format. The new method is called COPY_USE_MT. The data is read in chunks by CLI and the block is used for feeding a memory table. The memory table gets used in an INSERT..SELECT statement. The default chunk size that is used for INSERT...SELECT into target table is 64K rows. The new method gets applied automatically if moving row tables to columnar table format. If the table that is to be moved has (large object) LOB columns this approach cannot be used.
ADMIN_MOVE_TABLE now works with range-clustered tables. You can now use the organize_by_clause parameter of the ADMIN_MOVE_TABLE procedure to move range-clustered tables or change the organization form.
Pacemaker software refresh. Pacemaker has been upgraded from version 2.1.6 to 2.1.7, and crmsh has been dropped, as cibadmin is now used.
HADR_SSL_LABEL and HADR_SSL_HOST_VAL database configuration parameters are now supported in pureScale environment. See Configuring TLS for the communication between primary and standby HADR servers and Configuring hostname validation for encrypted communication between primary and standby hosts in an HADR environment for details on how to use these features.
Automated failover for DPF on Linux® for on-premises deployments with Pacemaker is now available for production environments. Pacemaker is replacing TSA as the supported cluster manager for Db2 on Linux in Db2 12.1. You can now deploy automated failover for DPF solution for on-premise instances with Pacemaker as the cluster manager. Users who are upgrading from earlier versions need to create their Pacemaker domain for DPF on Db2 12.1 using the db2cm command.
The following table lists backup and restore enhancements that are available in 12.1:
Table 2. Backup and restore enhancements in 12.1
Enhancement Description
Intra-table-space Parallelism (ITP) for backup operations. Intra-table-space Parallelism (ITP) enables Db2 to use multiple threads when backing up each table space. In cases where one or more table spaces are much larger than the others, the overall elapsed time to run the backup operation is substantially reduced.
Changes to LIST HISTORY, DB_HISTORY and db2HistoryData data structure. Five new fields are added to the history file entry: The LIST HISTORY command, DB_HISTORY administrative view and db2HistoryData data structures now accommodate these new fields. In addition, the maximum size of the COMMENT field is increased from 30 characters to 254 characters.
ZLIB compression for backup and log file compression. With the release of Db2 12.1, hardware accelerated backup and log file compression are now available for Linux systems equipped with PowerPC for Power9 or higher CPUs.
Automatic Db2 backup enhancements. With the release of Db2 12.1, automatic backups now support remote storage targets. In addition, automatic backups now support traditional automatic pruning for backup images.
For more details, refer to Automatic database backup.
New database configuration parameter for implicit, default row-modification-tracking-enabled schema creation. In previous releases of Db2, the row modification tracking (RMT) feature used for logical schema backup operations had to be set by the user during a CREATE SCHEMA or ALTER SCHEMA operation. With the release of Db2 12.1, a new database configuration parameter can be set to enable RMT for all schemas implicitly. For more information, see dft_schemas_rmt - Default schema RMT parameter.
New table function for generating details of a logical backup. A new table function, LOGICAL_BACKUP_DETAILS_TAB, now returns details of the logical backup images on a specified path or media. The details include timestamps for the backup images and the schemas and tables contained in these backups.
New Registry variable DB2_VALIDATE_LOG_ON_ARCHIVE. In Db2 12.1, a new registry variable DB2_VALIDATE_LOG_ON_ARCHIVE is available to specify whether Db2 performs validation before sending log files to archive. For more information, see Log file validation during archiving.
Offline backup requirement lifted, following a drop member topology change event. In Db2 pureScale® environments, after an offline drop member topology change event involving a cataloged database, an offline backup is no longer required. After restarting the Db2 instance, any first connect, including utility operations, can be driven on the database. This action updates the database topology to reflect the new member topology.
Support for database and table space restore involving dropped members in a Db2 pureScale environment. In Db2 pureScale environments, database or table space backup images can now be restored when the member topology in the instance differs between the backup image and the target database. This applies to both snapshot and non-snapshot recovery scenarios. For more information, see Backup, restore, and rollforward operations within Db2 pureScale Feature.
Support for database and table space rollforward through a drop member topology change event in a Db2 pureScale environment. In Db2 pureScale environments, a database or table space rollforward operation is now allowed through a drop member topology change event. For more information, see Restore and roll forward through a topology change.
Replication enhancement using the db2ReadLog API in a Db2 pureScale environment. In Db2 pureScale environments, replication using the db2ReadLog API can now read log record data from before a drop member topology change event.