why you don't play with ecap ?. it should faster than icap. greasySpoon based on java, i'm not surprised consume much memory. with i/e-cap you could also cache post request by using respmod vector. On Wed, Jun 13, 2012 at 11:31 PM, Eliezer Croitoru <eliezer@xxxxxxxxxxxx> wrote: > as i was working with ICAP i saw that GreasySpoon ICAP server consumes a lot > of memory and on load takes a lot of cpu from an unknown reasons so i was > looking for an alternative and didnt find one but i found a basic icap > server that i modified to be more modular and also to work with > instances\forks. > > the main goal of this specific modification is to make it simple to use for > url_rewriting. > > tests that was done until now for performance was on: > client--->squid\gw------->server > 1Gbit lan speed between all > client spec - intel atom 410D 2gb ram opensuse > squid spec - intel atom 510D 2GB ram Gentoo + squid 3.1.19 + ruby 1.9.3_p125 > server spec - 4GB core i3 opensuse 64 bit nginx serving simple html "it's > wokrs" > > with apache benchmark tools: > ab -c 1000 -n 4000 "http://otherdomain_to_rewrite/"; > > served all requests and about 800+ reqs per sec. > > download at: https://github.com/elico/squid-helpers/tree/master/echelon-mod > > looking for testers to make sure that the server is good. > > notes: the forks aren't build that good so in a case of termination by > exceptions runtime error only one fork goes down and you must kill all the > others manually to to restart the server. > > logs have a huge amount of output on a production environment so it's > recommended to not use it at all if you dont need it. > > > -- > Eliezer Croitoru > https://www1.ngtech.co.il > IT consulting for Nonprofit organizations > eliezer <at> ngtech.co.il
