On Fri, 24 Dec 1999, 8823559 wrote:
> I mean, since hit-ratio is specified in the server(robot)
> configuration, "cache_mem" seems not have effective influence on cache
> performance, does it? Then, how (if possible) to measure the effects
> caused by different settings of "cache_mem".
AFAIK, cache_mem might affect response time, but not hit ratio. You can
run several tests with identical workload, varying cache_mem (or any other
single parameter) to study its influence. Note, however, that Polygraph
offers "constant" hit ratio, and a proxy cannot cache more than that.
> And, did you ever use "cache_dir cachesize 1st-level 2nd-level"
> as a parameter of performance testing, especially the structure of
> the cache server? If you did, what were the results(or suggestions)?
This question is probably more appropriate for the squid-users mailing list
as it has nothing to do with Polygraph or benchmarking in general. Squid
performance tuning, including cache_dir parameters are often discussed on
squid-users; you may want to check the archives for recent suggestions.
This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:10 MDT