Hi everybody! My question may be rather theoretical, but in essence I need to know if Squid really has a flaw regarding latency for connections where keepalive is on. At ApacheCon 2014, Bryan Call presented slides where slides nr. 40 to 49 show where he writes on slide 46 about Squid: "Worst median latency for keep-alive benchmarks" . The slides are here: http://www.slideshare.net/bryan_call/choosing-a-proxy-server-apachecon-2014 The configuration for Squid is shown on slide nr. 36. To my eyes it looks a little over simplistic. I hope he has not configured Squid correctly and that somebody here can point me at better configuration that expressly does not have latency of many seconds and a 95 percentile of over 10 seconds. Those numbers were achieved by mesurement using CoAdvisor ( see http://coad.measurement-factory.com/cgi-bin/coad/FaqCgi?item_id=ALL ) My intent, is to use Squid with CARP or VRRP as a reverse proxy and load balancer for a cluster of webservers. My main reason for using Squid rather than NGINX or ATX or Varnish is Squid's superior protocol compliance. Byan Call's demostrated latency gives me reasons for concern. I spent the last weeks searching but I have not found anything that seems to counter Mr. Call's claim. On behalf of the Squid developers and users, I would be wery grateful if anybody could show or demonstrate the contrary. Preferably with configuration. About me: I have been a Squid proxy admin for almost 10 years now, and also administrating web cluster solutions for a small university. I am already deploying VRRP with NGINX as a load-balancer, but me and my coworkers are not satisfied with its performance. Best regards, -- Anna Jonna Armannsdottir <annaj@xxxxx> University of Iceland Computing Services _______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
