Quickly Checking If A Change Hurt Performance Using Stori

Posted on 31-Oct-2013 by Craig Shallahamer, craig@orapub.com

Quickly checking if a change hurt performance

Suppose a key module was changed and placed into QA or production. You need a way to on-the-fly analyze performance. You don't a bunch of numbers, a pile of Statspack pages, or to spend your afternoon doing drill-downs. What you need is an understandable summary with the option to delve deep into the details. This is a great use case for Stori!

Here's what to do...

First, create a Playback file containing the following Stori commands. (A Playback file is simply an flat file containing Stori commands.) I'll call the Playback file, quickcheck.pbf.

start statspack snap
pause 300
start statspack snap
set scope latest
analyze summary

Second, whenever you want check performance for a 300 second interval, run the Playback file. You can run the Playback file from the OS command line like this:

$ ./storiLin64 -pbf quickcheck.pbf

You can also start a Playback file from within Stori like this:

How can I help you? start pbf quickcheck.pbf

It's that simple!

Want to go a step further?

What made this example so straightforward is the use of the Playback Facility and the flexible analysis scope change. There is a lot more you can do with Playback files. For more information ask Stori, help pb or check out the Stori FAQ | Advanced Use | Can I script a sequence of commands?

Also key in this example is a quick and flexible analysis scope change. There are a number of ways to tell Stori to change the analysis scope. Notice I did not need to know Statspack the snap_ids or even the date! I simply asked Stori to reset the analysis scope for the latest two Statspack snapshots. It can't get any easier than that! For more information about changing the analysis scope ask Stori, help scope.

All the best in your Oracle database performance tuning work!


If you have any questions or comments, feel free to email me directly at craig at orapub.com.

CPU Cores Vs Threads: Utilization Difference..Part2B The Oracle Database Parallelization Sweet Spot Part 1 IOUG 2016 Is Different: Here's How To Get Your Abstract Accepted