Re: ICP_HIT_OBJ

From: Alex Rousskov <rousskov@dont-contact.us>
Date: Sat, 31 Jul 1999 15:31:46 -0600 (MDT)

Alexei,

        Seems like what you you may want to do is to finish/polish the
initial implementation of URNs in Squid. IIRC, current implementation
can already return a list of candidate sites (URLs) for a given URN so a
user can pick and choose. Adding response time estimations or ordering
locations by some measure of "goodness" would complete the algorithm. A
good implementation would also try to keep user in the URN space if
possible (something Squid does not do).

The tricky part is how to build the list of candidates, of course; but I
guess you already have more than enough ideas on how to do it.

Alex.

On Sat, 31 Jul 1999, Alexei Novikov wrote:

> Alex Rousskov wrote:
> >
> > On Fri, 30 Jul 1999, Alexei Novikov wrote:
> >
> > > I was wondering if it is still possible to use OP_HIT_OBJ as an opcode ?
> >
> > AFAIK, Squid no longer emits OP_HIT_OBJ. Returning content via ICP/UDP
> > is considered too dangerous and not worth the trouble. I would not
> > recommend supporting this operation if you have a choice.
> >
> > Alex.
> >
> > P.S. I am by no means an ICP guru though.
>
> Alex,
> what I wanted to do is to have a fake cache server that will sent a
> false hits for requests to some of the ftp sites providing a list of the
> nearest mirrors. There are generally 3 ways to do so:
> 1) to have a redirector at each cache server that will point to some
> (WWW) server that will give the list. Then a user will have a choise to
> get original content, and cache manager will have a greater flexibility
> (specifing domains that are regarded as far-away and as a nearby).
> 2) to use a fake cache server that will sent false hits and have a
> redirector of the siblings HTTP requests to the WWW server. In such a
> way users will have ability to reach original content.
> 3) to use a fake cache server that will sent false hits with the list
> inside the body.
>
> So I was just examining all these 3 ways.
> Alexei.
>
Received on Sat Jul 31 1999 - 15:12:58 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:47:39 MST