Re: [squid-users] Proactive refreshment in squid

Henrik Nordstrom wrote:
>>
>>Does squid support proactive refresment?

> What probably will be supported in a later Squid version is relaxed
> refreshes

I like that approach, and it sure makes sense. As of now how does Squid
handle concurrent requests for the same resource? This is a common
scenario for the setup I'm trying to put together (large sites with
Squid acting as a reverse proxy), and I was wondering whether, with no
object in cache, two concurrent requests would resolve in two separate
connections to the origin server (with some kind of cache arbitration
for subsequent writes, I'd assume) or there is some sort of smart queing.

> The problem with proactive refreshes is bandwidth management. All
> implementations in other cache servers known to the Squid developers
> tends to end up wasting very large amounts of bandwidth on objects
> which in the end is never accessed again by the clients, and until a
> reasonable approach to avoid this is found the Squid developers does
> not find any reason why to implement proactive refreshes in Squid.

Then again what about http acceleration? Normally bandwith isn't an
issue there, and I think users might benefit a lot from that. Also, I
understand that there is some sort of scoring algorithm with statystical
gathering, so I guess it should be feasible to proactively fetch just
frequently requestes objects.

Ciao,

-- 
Gianugo Rabellino
Pro-netics s.r.l. -  http://www.pro-netics.com
Orixo, the XML business alliance - http://www.orixo.com
     (Now blogging at: http://blogs.cocoondev.org/gianugo/)