Ian Holsman wrote:
Thanks... what I was after was more hints on configuring HTTP. ie.. make sure FollowSymLinks is On & AllowOverride off (to avoid unnessecary fileops) and things like this
Oh, OK, then http://httpd.apache.org/docs-2.0/misc/perf-tuning.html would be
relevant.
yeah.. I need to read it again, and see if I can add any perls
I set up my specweb99 DocumentRoot subtree to be as flat as possible, thinking that would cut down on cycles spent walking the directory paths. I can't say for certain that it makes a significant difference on Linux; the kernel has an efficient filepath cache.
Another area for potential improvement here is the use of file caching. I tried using mod_cache's fd caching, but quickly ran into problems with the Linux per-process fd limits because SPECWeb99 accesses so many different files. It's possible that caching the file contents in memory might allow more files to be cached. But any attempt to cache files in userland will tend to negate the kernel's file caching, which is pretty efficient on Linux.
hmm.. did you use a recent version of mod_mem_cache.. you should be able to set the maximum number of fd's to be under the limits (not sure if that would help.. depends on the caching)
The choice of MPMs makes a difference too. I get better results with prefork on Linux, with worker not too far behind. With a good userland file caching strategy, worker might do better if it has a higher cache hit ratio and lower overall RAM usage.
yes.. prefork with threading completly disabled helps
Greg
