On 20/02/2014 1:11 a.m., Dr.x wrote: > hi all , > i found that sites like aaaa.com as an example is the most site being viewd > by customers. > the question is how do i use squid with refresh pattern so that it cache > all that website for about 5 minutes and after that go agian cache it , and > so on . > > > if i do that im sure that byte ratio will be better It probably won't be. Or maybe the ratio will but the customers experience drop through the floor, so they move on elsewhere and you haev to start all over again. Begin by evaluating for each of the popular URLs: a) whether it's an ongoing or temporary trend. - optimizing closely for todays fad is often useless on tomorrows. This is why caches have dynamic timing heuristics and do revalidation. *Let it*. b) whether or not the resource/object is cacheable at all. - private, no-store, and no-transform exist for good reasons and are NOT usually used by default on web servers or frameworks. If they are present the author is making a clear statement about traffic safety. Touching is BAD. It actually wastes a lot of bandwidth and transfer time when authors having to re-override proxies which ignore those settings. c) Pay attention to the caching control headers. - If you see complicated settings chances are VERY high that someone is actively tuning it. So touching it yourself without *full* knowledge of the site and its behaviour patterns will break things. d) Whether forcing more caching for that site/service will bump something useful out of storage. - see (a). PS. remember that Squid can and *does* cache objects even when they are marked "no-cache" or with Authenticate: headers. No need for special overrides on those. HTH Amos
