Re: A stoopid question? (disk object balance)

From: Edwin Culp <>
Date: Wed, 14 Jan 1998 18:13:21 -0600

Cameron Blackwood wrote:
> G'day,
> I am looking at setting up a squid cache here and I found the page on
> sizing for intel boxes very informative. I noticed that many people on
> that page had 'N' caches load sharing and I am wondering just how to
> configure mine to load share (well not load share in a network sense but
> a disk sense). Do I have to hack something like pluggw to get the objects
> spread over two machines?
> Basically I intend to have:
your might reconfigure your DNS and put something like

proxy1 A .9.10
                A .9.11
proxy2 A .9.10
                A .9.11
proxy CNAME proxy1
proxy CNAME proxy2

Then connect them as siblings proxy-only to not duplicate the
use as the proxy not proxy1 or proxy2. This will
distribute alternate html's or ftp's.

> +------------+ +------------+
> | proxy 1 | | proxy 2 |
> | IP: .9.10 |---------| IP: .9.11 |
> | IP: .9.20 | | | <--- 9.20 if proxy1 fails
> +------------+ +------------+
Then to make it more interesting point everyone to another (third) squid
configuring "proxy" re:namedb above, as parent and it will parse
the html and more or less evenly distribute it between proxy1 and
proxy2. This also works to distribute load on a multihomed low
site with minimum resource demands.



P.S. anyone see a problem with this?
> with one box having a second IP number of 9.20 and 9.20 being the
> advertised IP address of the web proxy. .9.20 will fail over to the
> second box should the proxy1 fail. Both proxies will be siblings of
> each other, so from that point of view they will be 'one cache', but
> how do I share the requests so that the objects are shared over the
> machines.
> Now at this stage you are probably saying 'use a cisco redirector box'
> but I am too cheap for that :). I dont want to use a DNS magic
> solution either (we dont run smart DNS and its hard to change it). I dont
> really care about net bandwidth between these boxes (we have a
> switched 100M ethernet), so what I really want is to share the requests
> between the boxes directly so the objects end up on disk on each box.
> I cant see any way for squid on proxy1 to send, say 50%, of its misses
> to proxy 2. I think running a 'nanoproxy' on the 9.20 address that has
> both real caches as its parent and uses round-robin but no local
> disk space it self looks too hard (well it would be easy, but I cant
> see an easy way to do it without using more disk space :( and I just
> dont want that complexity).
> My solution is looking like writeing something like pluggw which sits on
> 9.20 and shares the load by redirecting the requests to 9.10 or 9.11 (randomly
> or something).
> I know I am going to get problems with object consistancy etc, but I
> can live with that.
> Any comments on this?
> cheers,
> cam
> --
> / `Rev Dr' skeptic, virtual goth \
> < [+61 3] 9669 4268 BSD Unix, C/C++, genetics, ATM >
> \ The BoM. /
> ____ finger for PGP/Geek Code and stuff ____
> On the side of the software box, in the "System Requirements" section,
> it said "Requires Windows 95 or better". So I installed Linux.
Received on Wed Jan 14 1998 - 16:12:52 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:38:25 MST