[Gloucs] books

[Gareth Bromley]

On Wed, 12 Nov 2003, Guy Edwards wrote:
> I'd like it so that if I had more than one squid server on a network,
> the load gets distributed between them, and if one dies the others keep
> going without anything stopping. I'd also like to be able to add another
> squid machine in at a later date and have it join in with the ones
> already present without having to start/stop configure much.
Lots of ways to do this based upon what your actually trying to do and
what budget you have.

Me personally would look at:
- SLB solutions to enable multiple chaces to act as one e.g. LVS, Foundry
ServerIron, Cisco CSS, Radware, F5 etc.. even good old Cisco routers :) if
your driver is just availability at one site.
- Caching hierchies using ICAP, HTCP or others to build up a tiered
architecture and as long as each child cache has 2+ parents have
resilience built in thta manner (Also helps keep content inside your
network(s) rather than leaking out to the internet and costing money)

> Is this how you'd do it or are there better ways?
If your messing with squid, make sure you have all the latest
patches/updates otherwise in 1000+ user envs you might find squid dies :(

Enjoy

G