[squid-users] Re: Why Squid is great (was: fourth cache off??)

----- Original Message -----
From: "Jon Kay" <jkay@pushcache.com>
To: "Joe Cooper" <joe@swelltech.com>
Cc: "khiz code" <khizcode@yahoo.com>; "fooler" <fooler@skyinet.net>;
<squid-users@squid-cache.org>
Sent: Thursday, December 20, 2001 1:04 AM
Subject: Why Squid is great (was: fourth cache off??)

> I think this is a good time to suggest a new doctrine on caching and
> cache placement. The way to get the most out of a cache is not to buy
> a really big one and centralize it, but get a bunch of small ones and
> put them topologically close to users.

true to push contents closer to the users.

> The main point of a cache is not to serve lots of requests, but rather
> to improve the network experience for a certain user population.
> Specifically, it reduces latency. Yes, it also saves on bandwidth,
> but the reason people care about that bandwidth savings is that users
> would have to wait longer if it were used by many requests for popular
> content.

request per second is only one of the factors of a cache.

>
> Now, the way a cache works best is if it is "near" its user
> population. A server delivers content more and more slowly not only
> when it is on the other side of a long WAN link, but also as the
> number of router and even bridge hops go up. The worst hops are, in
> general, at MAEs and other ISP boundaries, but even local router and
> bridge hops add their own slownesses, with store/forward latency, and,
> queueing latency, and lost packets which cause the worst slowdowns of
> all. That loss rate increases quasi-inverse-exponentially with hops,
> so keeping down hops is important even on LANs.

let the strong tcp/ip stack of a certain platform do its job.

>
> What does this have to do with Squid? Well, the best way to keep
> servers near clients, maximizing their effectiveness, is not to buy a
> really big machine room cache from which to serve the entire company,
> but rather to buy several, and place each near a different group.

this is true for siblings but what about their parents?

>
> The best way to do that is buy cheap caches with good latency. And
> that means Squid. The swelltech entry that we were all watching had
> latency almost as good as any of the Big Guys.

yes for a given request per second but increasing it will directly
proportional increase to other service times.

>
> So what can a Squid vendor do to sell to a T3-connected ISP? If
> the ISP has just or two offices and POPs, then you're toast. If they
> have several POPs, well, maybe now we're in business. Sell them one
> for each.

true if uptime is not important to them, if it is, then sell them at least
two and load balance it by an L4 switch or LVS

>
> Certainly, a big company, looking to improve effectiveness of even big,
> fat network connections, will be a possibility. Try to sell 'em one
> cache per group/remote office.

same answer as above.

> Of course, those of you in the Squid business have to get this out to
> your customers. Put at least a bit about it on your front page.
> Explain it in a white paper (I intend to do this for pushcache when
> timing is appropriate). Say something about it in the next cacheoff
> round. Mention it when big customers call.

the main reason why squid is great? it is simply because its free! but
performance point of view, still lots to catch... i must agree with joe
cooper's statement that its time to overhaul squid. :->

fooler.
Received on Thu Dec 20 2001 - 02:09:57 MST