Re: cache statistics

[Mark Nottingham <mnot@akamai.com>]

This is one of my favourite papers that touch on this subject, by
Craig Wills (though not much newer than the time frame you mention):
  http://www.cs.wpi.edu/~cew/papers/wcw99.pdf

I don't think there's necessarily a link between cacheability/etc and
the ability to serve a particular object via a CDN (if that's what
you're getting at), as objects are interpreted as "uncacheable"
(which in itself is a slippery concept) for a variety of reasons. 

I know this is frustrating, but there's no easy answer - you'd have
to look into why the pages have cookies, are marked uncacheable, etc,
and then see whether a CDN could provide the same services without
contacting the origin server; in many cases, it can.

Cheers


On Mon, Jan 08, 2001 at 07:57:21PM -0500, Renu Tewari wrote:
> Folks:
>    Does anyone have pointers to recent  statistics (later than 1998) 
> on what fraction of  web pages (and bytes) are static html and images.
> Similarly what fraction contain cookies/are explicitly marked 
> uncacheable /queries/dynamically generated/require database access. 
> 
> There was a claim made in a presentation that less than 10% of the data 
> (bytes) accessed on the web today is static html/images. Around 51% of 
> accessed web pages contain cookies and 48% are explicitly marked uncacheable 
> (with cache_control:private).  
> Are there any references available to corroborate or contradict this? 
> 
> 
> regards
> Renu
>  

-- 
Mark Nottingham, Research Scientist
Akamai Technologies (San Mateo, CA)