it means they didn't bother investigating the problem and reporting back to squid-users/squid-dev. They may find that Squid-2.7 (and my squid-2 fork) perform a ton better over whatever version they tried. I'm trying to continue benchmarking my local Squid-2 fork against simulated "lots of concurrent sessions" but the main problem is finding free/open tools to simulate "internet" traffic levels. Polygraph just can't simulate that many concurrent requests at a decent enough traffic rate without significant equipment investment. I have this nasty feeling I'm going to have invent my own.. 2c, Adrian 2009/5/2 Roy M. <setesting001@xxxxxxxxx>: > In http://highscalability.com/youtube-architecture , under "Serving > Thumbnails", it said: > > . > - Used squid (reverse proxy) in front of Apache. This worked for a > while, but as load increased performance eventually decreased. Went > from 300 requests/second to 20. > . > > So does it mean squid is not suitable for serving large ammount of > concurrent requests (as compare to apache) > > > Thanks. > >
