Re: Linux box

[Riccardo Losselli <riccardo@e4a.it>]

Il mer, 2002-07-31 alle 15:31, Jeff S Wheeler ha scritto:
> Riccardo,
> 
> You describe that you want all traffic originating from Net1 to traverse
> Router1, and traffic originating from Net2 to traverse Router2, in order
> to reach the Internet.  That is called policy-based routing, and you can
> implement it with iproute2 on Linux.
> 

Thanks, i'll look to find the documentation about it!

> You cannot both multihome using BGP, and policy-route in that manner. 

I don't want to.. the bgp approach is the evolution i would like to have
(btw, if i'm not wrong i can, with bpg, assign a priority to the routers
so that i can route 99% of the traffic to the "flat" one, and use the
pay-per use line only if the primary fails)
 
> In addition, although it seems like you have a firm understanding of
> what you want to do on this level, your organization probably lacks the
> necessary know-how to successfully deploy BGP, and your two ISPs may not
> even be staffed or equipped to deliver BGP sessions to you.  If you want
> to undertake it anyway, I strongly urge you to contract a consultant who
> can help you and possibly your ISPs through the process.

I know that one of the ISP can, and the other one could (i.e. he can,
but letting it understand what we need is quite hard... it took 1 month
to delegate the reverse resolution for our subnet, and when we asked to
allow us to monitor the router (the per-use one) with snmp it answered
it was not supported by the router, which is a cisco 2610).
My question was aimed to see if it is possible (and worth) for a small
ISP like us to implement bgp. Before doing that we will train the staff,
or use already trained tech personnel. I would also buy the routers
instead of renting it like now.. Also i've read many things about the
need to ask to have B classes assigned, or many C classes (we're in
europe, so we will have to ask RIPE). Leaving out all the problems about
the costs (i think that each class will cost, won't it?) it would be a
waste of addresses. Our actual need is no more than about 70 IPs, even a
full C net would be too much! 

Our main need is to avoid downtime and uneserray traffic between the two
nets, so i'm trying to see whats the best we can do. Do you think i
could successfully use policy based routing and DNS to have almost the
same results (i'm thinking about two nics on each server, one with an IP
of net1, the other with an IP of net2, and dns pointing normally to the
IP on net2 and switch to net1 if link2 goes down?) 
I know it will never be like bgp, but it still better than nothing at
all, or not?

Thanks again,
Riccardo