Performance testing for SAP BusinessObjects is somewhat of a touchy subject. Users want and need their data, and they want it fast. With so many high-speed technologies around these days, it is easy to understand why. Your BusinessObjects Admin can do a lot to help improve the performance of your reports, but that person can certainly use your help. Here are some tips of information that is useful for your Admin to help diagnose and fix a performance issue with your reports:
- Record the exact start and stop time of each report execution.
- Record the exact prompt values you passed (if any) so the administrator can re-run the report themselves. They might have a few diagnostic tools you don't.
- Don't run anything else. Run each report in singularity, and try to get the system as quiet as possible to avoid any interference with your test.
- Compile a nice big list with the Report Name, Folder Name (Admin might not know your structure), Universe Name if you know it (if not the Admin can get it), and the user name you used to run your tests.
- If you're looking at Operating System metrics, take a baseline first. You'd be surprised how many people skip that step. You have to know what your system looks like when no reports are running.
And just so you don't think I'm being a lazy admin, let me tell you why all of this stuff is important.
Nobody knows their data better than the developers. Administrators know your system inside and out, but generally don't have the best sense of the types of data or reports being run in there. Knowing what prompts your passed will help the Admin run the report the exact same way you did. Prompt values can have a HUGE impact on how many records are returned in a report. Is it fair to compare a 5,000 record report with a 65,000 record report? Nope.
Plus, your Admin likely has some sort of browser watch tool and can watch the trace logs in real-time if they can run a report again themselves.
Exact start and stop times help the Admin immensely when sifting through trace logs. If you saw how many files verbose logging can spit out in a minute… you'd likely need to take a seat before looking at that. Knowing where in that mess to start looking can REALLY help.
Performance testing is not an art. It's a science. It really has to be approached with discipline in order to be meaningful. By making sure the system is quiet or as quiet as possible, you'll increase the value of your performance metrics. If there are 500 schedule Webi reports firing off when you do one test, but not when you run the second test, is that a fair comparison? Nope. Apples to Apples. Oranges to Oranges. Each test must be as close to exactly the same as the next, or it is not a fair (or valid) comparison to another test.
The more information you can give your Administrator, the better they will be able to help you. The sooner you help them help you, the sooner everyone can get back to playing Angry Birds. 😉
