All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
apache+varnish slower than apache alone?
I want to speed up web-server (I have to use apache) running on my VPS, so I installed varnish. I did not play much with configuration, just moved apache to 8080-port (using mpm_prefork, in default 5/5/10/150/0 configuration) and varnish to 80-port.
I used "curl" to check if requests are routed properly (index.html is the default "welcome" one on debian, ~10kB static file):
curl -I http://127.0.0.1/index.html (there is X-Varnish header, so its goes over varnish)
curl -I http://127.0.0.1:8080/index.html (this goes directly to apache)
Now I'm doing some benchmarks, but to my surprise I get slightly better results for pure apache, than for apache+varnish:
ab -c 50 -n 100000 http://127.0.0.1/index.html (~2300 req/sec)
ab -c 50 -n 100000 http://127.0.0.1:8080/index.html (~2500 req/sec)
How is this possible? I expected varnish would speed web up by at least factor 2, especially for static cacheable files...
Comments
understanding what varnish is used for is important first before installing.
Unless you are bound by RAM. Doesn't varnish do in-ram caching?
Varnish does do in-ram caching, but in this case apache is taking advantage of system disk-cache (also in-ram). I have plenty of free ram right now, so it is clear to me I can not see any advantage here for static files.
What surprises me is how good apache handles many concurrent requests. I tried increasing concurrency to "-c 5000", still the same. I remember doing this test some 5 years ago, with apache being able to do barely 20% req/sec, compared to varnish+apache...
I'll do later some tests with my dynamic drupal-website, but for that some VCL-tweaks must be done first...