SSS-Test: SQL Scalability Simulation recipe
Load testing on a shoestring
Developers are given an assignment to quickly put together a new report to be run by 30 concurrent users. In unit testing, response time for a single user running this report is acceptable.
The question is, of course, if that will still hold true if the other 29 users jump in.
Any way to give a substantiated answer without actually asking 30 end-users to run this new report?
Perhaps an even more important question is whether slamming this new report on an already busy system might have any adverse affect of degrading system performance?
The new module is going through a unit testing, and you, as a DBA, are asked to evaluate any (hopefully none) potential adverse effects on the already existing and carefully tuned modules. What course of action would you suggest?
No new code path, same old well-known system, but now due to a recent merger there are expected to be at least 20% inrease in the number of concurrent users. You are asked whether your system can handle the load. How do you go about answering this question?
If the questions asked above don't sound unusual to you in your line of work you might find SSS-test of some value:
1/ "SQL Testing through Scalability with Respect to Concurrent User Count: A Simulation Recipe" paper (PDF)
3/ Additional resources, including hiperlinked html-based documentation, test results and appendixes mentioned in the documentation (after being redirected to the www.tns-0.com web site, follow the SSS-Test link)