Comments on: ArangoDB vs. CouchDB Benchmarking | ArangoDB 2012

By: Frank Celler

Frank Celler — Sun, 26 Aug 2012 10:13:00 +0000

In reply to jsteemann. Maybe one could use a simple pseudo-random number generator like "x * 13 + 5"?

By: jsteemann

jsteemann — Fri, 24 Aug 2012 16:57:00 +0000

In reply to Timo. There's so many things that could be tested, but unfortunately not so much time to do it. But yes, doing a cross-action benchmark would be nice and hopefully we'll find the time to do it at some later point. If so, I'd suggest not using random entries because it would introduce some non-deterministic behavior (which might or might not matter for the results).

By: Timo

Timo — Fri, 24 Aug 2012 16:27:00 +0000

I miss a cross action Benchmark, get, insert, update and delete at the same time.
Just one action at a moment is not a good test case, it should something like: I G I U G I D in a 100.000 loop. The Update and delete should target different random entries and not the last inserted.