IBM Z OMEGAMON AI for Db2 | Performance Database Streaming via OMEGAMON Data Provider

For as long as most mainframe shops can remember, batch reporting has been the reason customers turn to IBM Z OMEGAMON AI for Db2 when something needs explaining after the fact. Competitors praised their user interfaces for their look and feel, but when it comes to reconstructing what happened in a Db2 subsystem, which application consumed the CPU, which thread caused a locking conflict, who accessed what, OMEGAMON's batch reporter has been the tool of choice for decades.

That reporting engine draws several hundred Db2 performance metrics, organized into three broad categories:

• System level (STATISTICS report set) 
• Application level (ACCOUNTING and AUDIT report sets)
• Performance details (SQLACTIVTIY, IOACTIVITY and other report sets)

Accounting data tells you how an application, thread, or transaction performed: CPU, wait times, locking conflicts.

Dashboard: ACCOUNTING Report summary data for MAINPACK at Plan level - Commits/IO Wait time/Total Wait time

Dashboard: ACCOUNTING Report timeseries view for Top 5 MAINPACKs - Total Class 3 Wait time

Statistics data gives system programmers and DBAs the subsystem-wide view: data sets, memory use, SQL activity, commit rates.

Dashboard: SQL Open/Fetch/Close activity using STATISTICS Trace data

Dashboard: Aggregated Accounting Statistics (DDF connection type) using STATISTICS Trace data

Audit data, growing steadily in importance, tracks who did what with privileged access, who granted permissions to whom, and other security-relevant activity. SqlActivity reports allow you to follow the exact flow of an application by consuming very detailed performance trace records.

All data processed is drawn from SMF records or from OMEGAMON's own trace collection, which gives customers a way to gather this data independent of SMF.

What hasn't changed: Db2 on z/OS keeps adding metrics under its continuous delivery model, and OMEGAMON's batch reporting is typically first in line to support them, often from day one.

What has changed, as with OMEGAMON AI for Db2 6.1, is where that data can go. Through the OMEGAMON Data Provider, batch reporting output can now stream directly into modern analytics platforms, Elasticsearch, Splunk, Prometheus, and others, for visualization in tools like Grafana or Kibana. With one of the latest PTFs, UO07825, that coverage is complete: locking, exceptions, and index split tables are now included, meaning essentially every metric loadable into the Performance Database can also be streamed.

Dashboard: ACCOUNTING Report / Top 3 Nested Activities (Summary and time series) - Class 7 Elapsed time

This closes a gap that has existed since the OMEGAMON Data Provider first launched in 2021 and 2022. Online monitoring could only stream a limited set of system level metrics in near real time. Batch streaming removes that limitation, trading a small, configurable delay (every five or fifteen minutes, whatever fits) for the full metric set on a schedule customers control.

Just as important is what customers don't have to do: rewrite anything. The batch jobs that already build the Db2 load files that are used to populate the Performance Database keep running unchanged. A new, simple sample job provided in the TKO2SAMP library can be used to take the Performance Database FILE (Trace) or SAVE-CONVERT (Report) data as input and then invokes the necessary ODP API that will stream the data, through OMEGAMON Data Broker and Data Connect, to the desired target stack (for example Logstash/Elasticsearch).

Customized TKO2SAMP(FPEOSTRM) sample job to stream PDB data using ODP

The payoff shows up outside the mainframe team. Instead of writing SQL against a fifty-table Performance Database to pull security violations or trend data, auditors and site reliability engineers can work in the same graphical dashboards they use across the rest of the enterprise, drilling into time ranges, spotting trends, and correlating Db2 activity with everything else running in the business.

Decades of trusted batch reporting. Now built for the observability stack your whole enterprise already runs on.

#IBMZ #IBMChampion #AIOpsonZ #OMEGAMON