Re: What's the best configuration for this setup?

From: Ross Wheeler <[email protected]>
Date: Fri, 22 Nov 1996 18:57:45 +1100 (EST)

> we're trying to implement an ISP-wide caching project. most of the sys ads
> for the networks connected to the ISP chose squid.
>
> what's the best setup (neighbor/parent) for this config?
>
> o 64K link to local ISP (my network)
> o local ISP has 128K link to mother ISP
> o mother ISP has 512K link to MCI
>
> configuration for my network, configuration for local ISP and
> configuration for mother ISP.

On this subject, I'd like to ask a (dumb?) question.

It would appear, that "my network" above (64K) should use squid so users
get best us of that 64K link. Same can be said for each of the two next
systems upstream.

I have a (somewhat) similar configuration. I've 128K to one uplink, who
has 2Mbps to the 'net. I've a very small ISP downstream from me (only 3
modems), connected via 28K8 modem.

My upstream has squid. I have squid. The little fella at the end has squid.

I noticed a particularly unpleasant problem I'd appreciate ANY input on,
and that is that when someone requests a page through our proxy, and it
is NOT cached, it checks the "parent". IF upstream has it, (and here's
the catch), it comes roaring down the 128K and consumes most of our
bandwidth while it does so. If it's a large file (big web page, or worse,
a large file that's being FTPd), this can go on for quite some time.

Is there some (practical) way to "throttle" the fetch rate? It would be
really nice to set the maximum retrieval rate (per parent?) so if you are
on a limited bandwidth link, you can set it to some "acceptable" rate.
After all, if a user is only downloading at 28K8, there's no point in
sucking it down much faster than that... and leave the link free for
other work. If the neighbour is on say ethernet, then 1Mbps is more
reasonable, because it's less likely to impact on the local LAN at that
speed.

RossW
Received on Fri Nov 22 1996 - 00:00:13 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:33:36 MST