S-RAMP performance tests

brmeyer Nov 14, 2014 10:04 AM

I'm putting together a list of informal and formal performance tests, but would really appreciate your ideas on additional ways to hammer it. Thoughts, so far:

Load 1 million metadata-only artifacts and empirically test the responsiveness of the Java client, shell, and UI.
Load 1 million metadata-only artifacts and attach a profiler, identifying CPU hotspots and memory utilization.
Load several thousand document artifacts, attach a profiler, and check for hotspots and utilization within the processing and ArtifactBuilder concepts.
Repeatedly create and delete artifacts, profiling for memory leaks.

1. Re: S-RAMP performance tests

jorgemoralespou_2 Nov 14, 2014 12:34 PM (in response to brmeyer)

I guess that the queries should also be tested, just to also provide on best practices on how to create/relate/classify artifacts in terms of performance when querying.
I assume from your first tow points that each interface (java client, shell, ui) will be tested in isolation.

Also, important, I do not know if there is any clustering communication happening if you have 2 active S-Ramp servers, but if you do, very important to see behavior, in terms of normal performance and while failing over.
Actions