> I wouldn't call it advice, but we're thinking of trying iBGP because of
> the "multi-exit-discriminator" feature when we get a redundant link.
> BGP is also supposed to cope with both IPV4 and IPv6 routes with a
> single routing protocol.
Whilst I'm sure you realise this, it isn't explicit in what you say so
I'm going to mention it here as I have seen others wondering about
using BGP. BGP is usually run on top of an IGP, such as OSPF or
IS-IS. iBGP peerings are usually between the loopback addresses of
routers and each router needs to know how to get to the loopbacks of
the other routers.
For instance, within the JANET backbone all the point-to-point links
and loopback addresses of the routers are carried within IS-IS (and
only IS-IS), but all the site and regional network addresses are
carried within BGP.
Usually, an IGP determines how you get around your own network, and
BGP determines how you get to other networks.
Ian is correct about the IPv4 and IPv6 topologies needing to match
when using IS-IS, and that includes matching configuration at either
side of a point-to-point link. There is also multi-topology IS-IS,
but I don't have any experience of that.
Cheers,
Rob
|