Intro and outline

Hello, newbie onto the list, just introducing myself and indicating what
I'm working on.

I'm using squid heavily as an HTTP/S accelerator, and my current
development target is a load-balancing/failover backend for squid
supporting virtual hosts.

At this stage this revolves around specifying a list of machines which are
defined as back-end servers, these will have non-cached requests split
between them using a load-balancing algorithm (currently round-robin but
it has been split off into a seperate .c file so that other more advanced
algorthims, one of which I have in the works, can be plugged in).

I am currently working on the failover part, which will mark a backend
target unavailable if a connection is dropped, and retry it at a later
point. I hope to be able to advance this at a later stage to have
fine-grained functionality testing.

I am currently building this on the 2.3stable4 source, because I need to
use it pretty much now :) What is the opinion of the list on the stability
of the devel trees for acceleration at this stage? would it be
advantageous for me to work on that tree instead, or is a forward port
(theres not a lot of code really, its just working out the right places to
put everything :) likely to be easy enough?

The other question is whether you guys think this is appropriate for squid
itself. I'm doing it no matter what, I need it :) But if you don't think
its appropriate for the engine itself I can maintain it as a seperate
patch or something.

Anyway, hi everyone :)

Richard.
Received on Mon Oct 02 2000 - 17:11:56 MDT