Norbert Hoeller wrote:
I implemented squid 3.0 on a home system largely to help reduce Internet traffic (I am forced to use a 3G/cell-based connection that has high excess bandwidth charges). I used the default configuration - the only major changes were defining a 4GB cache and adding ACLs to block ad sites.
Squid appears to be running well (no complaints from users). I am surprised at the high "Hits as % of bytes sent" - I often see 60 minute values in the 10-20% range and I have seen higher values. I would have expected relatively little overlap between destinations visited by the users, most users are only using one browser (IE8 or Firefox 3) and they rarely clear their browser cache. I definitely see the same user requesting the same object (a JPG) and having squid deliver it from cache.
I never argue with success, but would like to understand how squid appears to be so much better than the caching provided by the browser.
Thanks, Norbert
PS. Great work not only on the squid code but also on the extensive documentation!
Not certain. But I suspect it has something to do with small browser
caches. The defaults I've seen on client boxes are usually in the MB
ranges. Large enough for a few dozen flash media videos and little else.
Your 4GB in Squid could be making up the difference.
10%->20% is a bit low. The ideal range is 40%+ for regular proxies. 10%
is on par with ISP using non-optimal object caching or without enough
cache storage to meet their users traffic load.
Amos
--
Please be using
Current Stable Squid 2.7.STABLE7 or 3.0.STABLE20
Current Beta Squid 3.1.0.15
