I'm trying to run some benchmark measurements by my own. The idea is to use some well known vanilla web benchmark (e.g. RUBiS, TPC-W) and their tuned version (Varnish reverse proxy in front Apache and software cache in front database). I have couple dilemmas:
Varnish should be 'relatively easy' to install and configure. However I'm more concerned about cache. If I understood correctly, software cache (e.g. Memcached, Redis) requires code changes. It means I need to change each database command in benchmark (potentially large number of servlets) and to use Memcached API. Is this correct?
Second, I need this results rather soon, so is there any alternative (some software cache which doesn't require code changes, or benchmark with already incorporate Memcached API)?
Thanks.