“Mainframes are expensive, and we need to make the best possible use of the processing power we have. We recently did an evaluation of competing products and we just could not find one tool that can do everything that TUC can. Total control of utilities is truly achieved only when you have an interface with the scheduler and integration with application processes.” — Yaron Zehavi
Discount Bank is one of the largest banking groups in Israel. In addition to its traditional branch system, Discount Bank is leading the retail market with new and innovative concept branches, internet and call-center based banking, and offers extended evening branch hours.
"We do not really have a batch window" explained Mrs. Tamar Tchelet, the head of the database administration team, "applications are running in a 24x7 environment and we need to be able to run our backups and reorgs without affecting application availability". TUC generates and executes all the utility jobs with full integration with the application batch flow. "We run all our utility jobs with TUC, including backups, statistics collections, reorgs, discards, checks, recovery and even copying data from production to test with DSN1COPY" described Tchelet, "and we allow the application jobs to trigger backups at significant business points in time."
Discount Bank stores all its business critical data in DB2 tables. Data sharing allows working in 2 live production sites ensuring full availability and recoverability in case of disaster. Batch jobs are running on multiple machines while workload is balanced by WLM. TUC allows a high level of parallelism and fully exploits partitioning. Jobs are handed over to CONTROL-M for scheduling and WLM to decide which machine best fits the current workload.
The bank converted its legacy applications from UNISYS to Z/OS in a five year project and implemented a state-of-the-art infrastructure to support the application development and production environments. TUC naturally took part in this implementation project, allowing the database administration team to drive the maintenance tasks hands free.
"We decided to implement TUC because it provided unique features that allowed us to save costs and reduce the number of jobs we had to run for backups and reorgs" explained Mr. Yaron Zehavi, the head of the database and software engineering group.
TUC is using Real Time Statistics to identify the rate of changes since the last copy. In addition, you can force full backups periodically using a calendar. "Before implementing TUC, we couldn't really take incremental copies and we had to take full copies", explained Tchelet. "It was too expensive and took too long. We found out that less than 10 percent of the data is changed daily so the difference between full copies and incremental copies is huge". TUC allows you to dramatically decrease the cost and the elapsed time for backups and backup tables based on the rate of changes using incremental copies. "We allow TUC to examine twice a day which objects require backup using Real Time Statistics. This way we can meet recovery SLA for critical tables updated frequently"
Backup Strategy
- Incremental copies taken to DASD twice a day based on RTS rate of changes
- Copy and QUIESCE ordered as part of application daily batch flow
- Copy policy rules enforced at the end of the batch window to ensure recoverability
- Full copy enforced once a week
- Full copy taken monthly using a calendar including all objects per regulations requirements
- Large Non Partitioned Indexes copied to speed up recovery
- Multiple COPY vendors are in use (IBM and BMC)
A common practice is to collect statistics from time to time to ensure the DB2 optimizer has the most current statistics required to choose the best access path for better performance. However, collecting statistics blindly may be expensive if the data has not changed dramatically or has not grown. Real Time Statistics can be used to identify the rate of changes since the last RUNSTATS. TUC exploits RTS and will trigger statistics collection only if the statistics are indeed outdated. "TUC allows us to collect statistics more efficiently" stated Tchelet. "It does not make sense to collect statistics if the data has not grown at least 20 percent. TUC collects statistics only for tables that really need it". Many availability disruptions can be avoided by taking preventive measures to avoid expected problems. For example, you can track the growth of tables, to avoid running out of space, or track tables to avoid pending states.
TUC automates the tracking processes and generates the preventive or corrective actions required to resolve conditions such as RECOVER, REBUILD, REORG, CHECK DELETE or START. "We are using TUC to automatically handle tablespaces in CHECK pending state." described Tchelet. "CONTROL-O can capture abends on check-pending and automatically submits a TUC job that creates exceptions tables and executes CHECK DATA with the DELETE YES option. Before we automated this process with TUC we got calls in the middle of the night and now everything is handled automatically by TUC".
Small tables are easier to maintain and access. You can decrease the size of your tables by archiving or purging old data. TUC automates the generation of the REORG DISCARD utility and allows you to easily purge or archive old data from your DB2 tables. This way you can trim down critical tables and improve access performance. TUC records the discarded flat files and allows you to easily reload or recall the data when needed. Applications can trigger Reorg DISCARD and set the required discard conditions. "Before introducing new tables to production, we always try to define within TUC discard conditions to purge obsolete data or archive historical data. This way we avoid performance problems with large tables" explained Tchelet.
Reorg DISCARD
- Datasets are recorded for recall so discarded data can be easily reloaded
- DB2 version 8 allows running REORG DISCARD in SHRLEVEL CHANGE
- Generating CHECK DATA for dependent tables with DELETE YES
- Creating and dropping CHECK DATA exception tables automatically
“Mainframes are expensive, and we need to make the best possible use of the processing power we have,” Zehavi explained. “We recently did an evaluation of competing products and we just could not find one tool that can do everything that TUC can. Total control of utilities is truly achieved only when you have an interface with the scheduler and integration with application processes”.
“When we compare Total Utility Control with other products, it continues to come out ahead in terms of features, ease of use, cost and performance,” Zehavi concluded. “TUC allows us to provide quality service for our clients, and position us to support the bank’s growth.”
