[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: max requests a celeron web server can handle

> I have another question - what is the optimal max. keep-alive time?
Because as I
> can see from Apache's /server-status page on our server, there are
usually about
> 10-15 processes in state "S" (Sending Reply) and another 40-50 in "K"
> (keepalive). In have lowered this time from 15 seconds to 10. Is there
> optimal setting?

Alternatively, turn keep alive off completely. Having keep alive on means
that if a person then clicks on something on the webpage within 10 (or 15)
seconds then it will load quickly as it doesn't need to spawn new children
to handle it. However each visitor can tie up 5 or 10 such processes, so
you can imagine the problem when there are many concurrent visitors. We
just turn it off, and it works well.

> What is the default browser behavior? I load www.somewebsite.com, with
20 little
> images. Browser makes few connections to server, fetches all images and
> connections stay in keepalive state. When I click some link on that web
> does the browser try to verify all the (same) images again? Or just
fetches the
> new page and maybe some new image?

Depends on their cache settings, if there is a proxy in front of the
person (many ISPs have transparent proxies), and whether you set "pragma:
no cache"... but usually the web browser tries to get the least amount of
stuff if possible.

Reply to: