David Young wrote:
I can already run ospfd on NetBSD without using MPLS or ldpd. What advantages does MPLS have over what I can do today?
In NetBSD now ? Probably none. Maybe yes, only if you have an already MPLS enabled network and want to take some load from your routers (this is also doubtable). Also maybe if you want to control the exit point of packets with a certain destination from MPLS domain in case your NetBSD is edge LSR.
I believe that MPLS may be useful for a pet project, but I do not know if I could justify adding MPLS to the kernel.
First of all I don't think the overhead is that much. Also there will be no MPLS code in your kernel unless you want too. I don't plan to enable it in GENERIC, INSTALL etc. I think the patch has options MPLS in GENERIC but this will be fixed.
> How do you think it is justified?I think it's justified for future developments in this area that I want to happen in NetBSD. I'm pretty interested to continue the work in this area.
or just setup them manually.How do I do that? Can you give some examples?
Currently, ldpd is set to listen on a PF_ROUTE socket for intersting routes and first of all, advertise to its neighbours about prefixes added/removed and second of all if it's necessary changes that route to point to an MPLS neighbour. So, in order to answer your question, adding normal IPv4 routes should be enough. This is not very smart for the moment but I hope to integrate ldpd with openbgpd sooner or later and get rid of this behaviour.
-- Thanks, Mihai