On Fri, 24 Jan 1997 06:34:40 -0800 (PST), Paul Goyette  
<paul@pgoyette.bdt.com> wrote:
> On Fri, 24 Jan 1997, Paul Goyette wrote:
>
> > But, what is really being asked for here is _policy-based_ routing, where
> > some policy dictates who gets to use which network resources.  This is
> > definitely not implemented in NetBSD, and is only available (AFAIK) in
> > high end routing products (like Cisco...), and only in recent versions.

Btw: "where some policy dictates who gets to use which network resouces" can  
just as well mean modification of the route processor scheduling algorithm,  
based on arbitrary priority assignment.  Something almost any router claims to  
be able to do, but doesn't neccessarily do terribly well.

> Actually, you don't really need a high end router for policy-based
> routing, all you need is to implement BGP4.  This is the routing protocol
> that all the big Internet boys use to exchange routing information with
> each other.

Huh?  What does BGP have to do with all this?  BGP standards for Border  
Gateway Protocol, which is used for interdomain routing, in other words  
disseminating information to insure loop-free routing to networks which are  
outside your AS.

You have one gateway, with two outbound links.  And you want to make a  
routing decision based on source.  You cannot, because you don't have a  
routing mechanism in the kernel to do this, accomplish that.

No matter what routing protocol you use to determine your gateways to other  
networks.  BGP has absolutely nothing to do with source routing based on  
policy.  BGP will certainly help you to decide how to get from your AS through  
which link to another AS to a final destination AS.   However, it will not  
provide you with the actual routing *mechanism*.

Back to my lovely European roast coffee.

Best regards,
Chris

--
Christian Kuhtz <ckuhtz@paranet.com>
                                                          ".com is a mistake."