mod_deflate at Perlmonks.org

Hi Fellow Monks,

Has Perlmonks ever tested the waters with mod_deflate? Since that vast majority of what PerlMonks puts out is HTML, from a bandwidth standpoint, one would expect a significant gain. On the other side, it does require more processing time.

So, I was just curious if it was ever tried here, and if so, what the results were.

Ted Young

($$<<$$=>$$<=>$$<=$$>>$$) always returns 1. :-)

Comment on mod_deflate at Perlmonks.org Download Code

Replies are listed 'Best First'.
Re: mod_deflate at Perlmonks.org by tye (Sage) on Feb 14, 2007 at 04:54 UTC
We haven't noticed a problem with bandwidth. We have noticed problems with CPU exhaustion for various reasons. So trying mod_deflate would mostly just be a way to make things worse, so we haven't really considered trying it. It could have an advantage for monks with slow internet connections, but given the time required for typical page loads here I don't think even that improvement would be easy to notice. - tye	[reply]
Re^2: mod_deflate at Perlmonks.org by TedYoung (Deacon) on Feb 14, 2007 at 13:31 UTC
It could have an advantage for monks with slow internet connections... I always wondered if mod_deflate helped modem users any. I wanted to try it once, but the only modem user I knew didn't have the time to test. I suspect it won't since PPP (which is done over the slowest link in the chain, includes compression. Ted Young `($$<<$$=>$$<=>$$<=$$>>$$) always returns 1. :-)`	[reply] [d/l]
Re^3: mod_deflate at Perlmonks.org by CountZero (Bishop) on Feb 14, 2007 at 20:23 UTC
Many, many, many years ago (far longer than I care to admit) I had a modem which did compression as part of the protocol it used. Whether it was a general applicable thing or only working between modems of the same type, I cannot remember, but it meant there was almost no benefit in sending compressed files. Of course that was before WWW was ever invented ... CountZero "If you have four groups working on a compiler, you'll get a 4-pass compiler." - Conway's Law	[reply]