Ulf Heinrich – Segus
Abstract
Most IT experts know the catchphrase “Never change a running system” very well, especially once you experience that a small change can have a significant impact. Anyway, “The only constant is change” is very true as well. Even if we don’t touch an application for enhancements, we’ll face security updates, software updates, hardware updates, configuration changes and data growth.
Before moving to a new machine, applying (Db2) APARs, changing schema, or SQL, it’s common to test in a sand-box, or development system – very good, but here come the challenges:
1st – There are quite a few regulations that affect reliable testing of changes and they all have one in common – DON’T USE PRODUCTION DATA!!!
2nd – Without real data, or with data masked and sanitized, we’ll very likely see different results than running one and the same query in production.
This session explains how you can reliably, and resource-friendly, simulate your entire production environment, WITHOUT production data and even on a different machine.
We’ll cover prescreening changes in the following areas:
- environment changes, like upgrading to a new machine, or adding more resources
- application changes, like SQL queries and schema
- data changes, like amount and distribution
- The capabilities introduce a whole new world of “what if” opportunities and especially protect you from testing that doesn’t focus on what counts: the resulting performance in your production environment.
Outline:
- What the Db2 optimizer needs to make a choice.
- How to model a production environment.
- Compare access paths resulting from buffer pool changes, RID expansion, or CPU changes.
- Real world examples.
- Test “what if” you drop, create or change an index.