Re: Squid memory footprint (was: MemPools rewrite)

From: Joe Cooper <joe@dont-contact.us>
Date: Mon, 06 Nov 2000 06:03:25 -0600

Andres Kroonmaa wrote:
>
> On 5 Nov 2000, at 11:53, Chemolli Francesco (USI) <ChemolliF@GruppoCredit.it> wrote:
>
> > WRT [1]:
> > Besides, if we get an ICP query and we hit, there is a reasonable
> > chance that we will get shortly a full-fledged query for that same
> > object. So keeping an in-core cache of the most recently
> > ICP-queried objects will save us a fair percentage of the overhead.
>
> The case when http fetch follows ICP-hit isn't a case for worry.
> The problem is for cases when fetch doesn't follow, like ICP-miss.
>
> Imagine a setup: 10 caches, loadbalanced with L4 switch, seen as a
> single cache by clients. Each box takes 1/10th of the http load, but
> to avoid cache pollution when each cache contains about the same
> content, peering setup is needed between the caches, and ICP gives
> currently best results. Thus all caches will see 100% of request
> rate as ICP queries.
> So, if ICP is slow, you either loose the benefits of loadbalancing,
> as ICP servicing will dominate the cache load, or have to accept
> non-ICP setup where all caches have eventually about the same content.
> Hard choice.
>
> In cache clusters we have to address icp (as a term, be it Digests,
> ICP or something new), that allows us in a fast and lightweight way
> to determine if any peer has requested object. Current Digests is
> not a good solution, during 1 hour between digest exchanges, lots of
> potential hits may get routed to the "wrong" cache.
>
> If we can keep ICP lightweight by some "microdigests" of squidFS
> metadata, and this microdigest can be updated instantly, then we
> should be ok.

I'm confused. Are there any L4 switches that don't provide a means to
schedule cache usage based on a hashed value or some other network
division to avoid redundant data? Even LVS now has a hash based
scheduling algorithm so that there is never any redundant data on the
balanced caches.

ICP shouldn't come into the equation at all in an L4 balanced network,
unless there are L4 switches that can't intelligently select a cache?

ICP has some good uses, but use when balancing caches is unwise, I
think.
                                  --
                     Joe Cooper <joe@swelltech.com>
                 Affordable Web Caching Proxy Appliances
                        http://www.swelltech.com
Received on Mon Nov 06 2000 - 04:56:33 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:12:55 MST