RE: HTTP serving compressed content (was Re: Perl "competitor" Cu rl raises $52M)

From: Damien Morton (Morton@dennisinter.com)
Date: Thu Mar 15 2001 - 16:15:51 PST


> > I havent noticed it in widespread use either, and I think
> the reasons are simple.
> >
> > Firstly, its probably too processor intensive to run on
> dynamic content (e.g request generated HTML pages).
>
> Not really -- I don't have specific figures to hand, but the
> processing
> cost tends to be fairly insignificant compared with the bandwidth
> saving.

I'll have to do some testing on this. I had assumed it would be significant.

> > Secondly, its not really much use when applied to the most
> intensive static
> > content (GIFs, JPEGs)
>
> No indeed, but there are plenty of sites that re-use the same graphics
> from page to page changing only the text -- news sites being
> an obvious case in point.

Right, but theres little to be gained from gzipping a jpeg, although even a
10% gain from a one-off compression isnt to be sneezed at.

> > Thirdly, it doesnt benefit modem users (on the
> ISP->modem->user leg) at all
> > (they already have compression)
>
> It does benefit modem users quite considerably because it tends to
> compress a lot more than modems do (we've seen 6:1
> compression ratios on
> typical HTML content).

Ive seen 4:1 and 6:1 figures bandied around for modem compession.

> > On the other hand, Ive been finding that rich media static
> content benefits
> > muchly (it seems to work a treat on certain kinds of Flash files).
> >
> > It would probably be most usefull implemented in a proxy cache.
>
> mod_gzip works pretty transparently with Apache.

IIS caches the compressed content, how does mod_gzip work?

> I can't actually understand why more people aren't using
> compression on
> their servers -- in the right circumstances it's extremely effective.

No-one ever won a pissing competition through conservation.

But it can cut your bandwidth costs :)



This archive was generated by hypermail 2b29 : Fri Apr 27 2001 - 23:14:15 PDT