AIOps on IBM Z - Group home

DB2 tracing support in Instana for z/OS

  

DB2 for z/OS Distributed Data Facility (DDF) is growing in popularity for multiple reasons

1.       Reliability, Availability and Serviceability of IBM zSystems and z/OS in particular

2.       Security on IBM zSystems

3.       Hybrid cloud computing model

With ever growing hybrid cloud computing popularity, customers tend to keep their applications on the cloud, but prefer to keep their database on-prem. Considering the increased customer adoption of this model, it becomes imperative Instana for z/OS has the ability to trace the DDF flows into DB2 . To begin with, Instana for z/OS released a new feature that enables tracing a JDBC flow into DB2 for z/OS.

The salient features of the design are:

1.       Makes use of SMF accounting 101 records for the performance data.

2.       Makes use of SMF statistics 102 records for deadlock and timeout data.

3.       Common Data Provider (CDP) is used to stream the DB2 SMF records to the Distributed Gateway (DG), part of Instana for z/OS.

4.       The DG transforms the SMF records to spans and forwards it to the Instana backend.

5.       Instana for z/OS is not in the critical path of the transaction.

The data provided in the first release include

·         CPU time

·         Elapsed time

·         Lock time

·         IO wait time

·         Plan name

·         Deadlock - blocker and waiter

·         Timeout - blocker and the timeout thread

To trace JDBC calls into DB2 for z/OS, SMF records are used. So, some of the DB2 settings/requirements are

1.       DB2 Accounting trace should be started for Class 1, 2 and 3

2.       DB2 Statistics trace should be started for Class 3

3.       ACCUMACC should be set to NO, so that performance records are cut for each unit of work

In summary, the JDBC flow into DB2 for z/OS is the initial flow that Instana for z/OS will be providing tracing support for. More to come…So, watch this space.

For more information on AIOps for IBM zSystems solutions visit:  https://ibm.biz/New-AIOps