John did the benchmarking but basically yes. We can absolutely hammered the server with Squid running and it barely touched the processor. The thing to remember is that Squid caches the final HTML so for pages that are not personalised (almost all in our case) there are virtually NO hits to Midgard/MySQL at all. If you set your expiry for say 15 minutes, then each page will only get hit once every 15 minutes. Then when pages are hit, it uses the last_modified and etag headers to do quick lookups to check the last updated date of the midgard article/topic/page and then if the page hasn't changed, it sends a 304 response telling the proxy server to server the one out of cache for another 15 minutes. So it not only takes 90% or more of the work away from Midgard/PHP/MySQL, but when it does get hit, it only processes the full page when it is new and needs to be updated.
Cheers James
Tony Lee wrote:
James,
Assuming your whole site isn't based on personalisation, implementing Squid as a web accelerator with proper caching headers on your pages can allow Midgard to scale to unforeseen heights.
Wow! Did I read right? The squid setup has a better response time for 70 users than Midgard at 10?
Tony
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
