> Hi! > > On Tue, Apr 28, 2009 at 7:54 AM, Wilson Hernandez - MSD, S. A. > <w...@msdrd.com> wrote: >> Hello everybody. >> >> I am somewhat confused on how squid helps to save bandwidth. I know it >> saves visited websites to cache and when someone else request the same >> site it will serve it from the cache. Please correct me if that is >> wrong. > > Yes, but I think that it "verifies" if the page is outdated (hence the > request).
Correct. These are seen as 3xx responses and IMS_HIT or REFRESH_HIT in the logs. Bandwidth the size of the headers is spent in order to save additional bandwidth to the size of object being requested. > >> >> Now, I've been checking my traffic before (external nic) and after >> (inside network) squid. Eveerytime I request a page (google.com) the >> request is sent to the internet as well so, in this case there isn't >> much saving done. But, if I have offline_mode on I get the old page >> stored locally. It seems that if I have offline_mode enabled is when >> bandwidth is been saved. > > In offline mode, I think it no longer verifies if the pages are > outdated. Also, due to the "interactive" web pages these days, there > are lots of sites which include the "no-cache" directive on the > headers, and squid have to honor these..... > > You should also verify the maximum and minimum object size, in order > to tune the cache for your particular case. > > Please somebody correct me if I'm wrong. You are correct. Google.com being one such websites which explicitly specify their page must be reloaded after 0 seconds is a particularly bad test of caching. Use http://www.ircache.net/cgi-bin/cacheability.py to see what pages are able to be saved, and why if not. FWIW: offline mode assumes that no bandwidth can be expended to update information. A lot of the web simply breaks, some well designed parts remain working for a time. Amos
