On Thursday 10 April 2008 12:26:57 Andy wrote: > It's only safe to cache data when you know it is cacheable.
I agree with many of the sentiments you raise, but a key thing to remember is that many caches will not cache objects over a certain size. Lots of heuristics around caching have been worked on over the years, but most of them generally work with objects within certain general parameters. (I worked in the caching industry on scaling delivery both on content sending and content recieving ends for several years, and could really go on about this, but I won't :-) Sure you could whitelist specific sites, but also caching in the ISP doesn't deal with the problem people have been raising - as previously mentioned in this thread: On 09/04/2008, David McBride <[EMAIL PROTECTED]> wrote: > The ISPs who are anticipating financial hardship are more concerned with the > cost of bandwidth between their network and home ADSL users, and _not_ > between their network and the outside world. And whilst setting the content to be cacheable is a nice idea, I doubt that ISPs are likely to want to maintain whitelists or to increase their maximum object size that high. But then its this sort of niche that companies like akamai have been targetting for years now. I can see companies liking the idea, but I'm not convinced companies really want their staff watching doctor who or similar at work... If the BBC wanted to build out their own CDN system (debateable/no idea), then personally I'd think a system based around L5 multicast ideas combined with content caches at the edge - like scattercast - is probably one of the best approaches. But then I've been saying that for years now since I've seen that work well in the past ... :) That said, it still doesn't help with the problem David mentions above. Regards, Michael. - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/[email protected]/
