Db2 for z/OS Monitoring and Performance Management

Health Insights Assessments – Focused Views

Numerous drilldowns are available to move from a high-level enterprise-wide view into focused analysis to help identify actionable insights for specific Db2 members, buffer pools, etc. The following image captures the outcome of two such drilldown steps, initially “Pool by Size” (not shown, indicating these exceptions are occurring in buffer pools belonging to the set of 4K buffers) and then by “Buffer Pool” to isolate the exceptions to specific buffer pools (shown here).

Health Insights Assessments – Time Charts

At any phase of the analytical process “Time Charts” of all assessed metrics can be generated to examine potential high-level relationships between the metrics. In this example, possible time-of-day correlations between the two metrics with exceptions (with the orange and red borders) and overall getpage activity (in the first chart) can be evaluated, and any specific metric can be selected for more detailed analysis.

Extensive Visibility into Buffer Pool and I/O Metrics

Since Db2 relies on having needed data residing in a buffer to avoid I/Os that are synchronous with the unit of work (“random sync read I/Os”), extensive visibility into buffer pool and I/O metrics is crucial for Db2 performance tuning.

Seamlessly Integrate Statistics and Accounting Data

Context-sensitive drilldowns are especially powerful when working with large quantities of data, enabling you to zero in on the subset of data pertinent to the specific type of analysis currently being performed. As one example, industry Db2 performance experts frequently suggest analyses focused by connection type, since online work (e.g., coming into Db2 from CICS) typically has a very different profile from batch work (e.g., coming in through IMS batch BMPs).

In this example, a drilldown seamlessly integrates connection type (from Accounting data) with sync read I/Os (from Statistics data) to create this time of day profile.

Integrated Visibility into Db2 I/O Statistics by Buffer Pool and Database

Powerful insights into disk I/O and cache performance by Db2 buffer pool and database can be facilitated by integrating data set I/O performance data (from SMF 42 records) with Db2 dataset I/O statistics (IFCID 199) data.

Metrics including disk response time by components (IOSQ, Pend, Disc, Conn) and disk cache hits and misses are presented, by Db2 buffer pool (along the top row) and database (along the bottom). And any of these can be selected for more detailed analysis, leveraging numerous drill-downs.

Visibility into Db2 Elapsed Time Profiles

In Db2 accounting data, combining “class 2” (CPU) and “class 3” (wait) times provides an elapsed time profile of time spent within Db2.

Here we see such a profile for work coming from CICS, showing that database sync I/O time (in yellow), CPU time (in red), and Global Contention for L-Locks (gray) are the primary contributors to elapsed time.

Drilldowns by CICS Transaction ID

For work coming into CICS from Db2, we can take advantage of the fact that the calling transaction ID is included in the correlation name field found in the Db2 accounting data to facilitate numerous types of analyses by CICS transaction. This example presents a view of Db2 elapsed time profiles by CICS transaction.

Integration and Visual Comparison of CICS and Db2 SMF Data

Also capitalizing on the presence of the CICS transaction ID in the Db2 correlation name field, additional integration of the measurement data produced by the two subsystems is provided. In this example, CICS SMF 110.1 transaction data (on the top row) is combined with Db2 SMF 101 accounting data (on the bottom row). This facilitates visual comparisons between the two data sources of activity rates, CPU, response/elapsed times, and activity rates by CICS region/Db2 member.