Re: [rrg] Rebooting the RRG

[Scott Brim <[e-mail]>]

HI Christian.

On Fri, Nov 15, 2013 at 5:29 PM, Christian Esteve Rothenberg
<[e-mail]> wrote:
>
> I would like to see RRG embracing work on two areas:
>
> 1.- Within the ICN routing paradigm (in the NDN project), hyperbolic routing (by UCSD/CAIDA, Dmitri Krioukov, Ken Keys, and et al.)
> http://www.caida.org/publications/presentations/2012/hyperbolic_routing_ndn/hyperbolic_routing_ndn.pdf

Yes, I think ICN routing is a fertile field as well, growing lots of
interesting flowers. I wouldn't want to commit RRG to endorsing
hyperbolic routing immediately. There are useful questions that could
be discussed, for example the constraints of real world business
relationships on greedy forwarding. We've tried related ideas in the
past (coordinate-based routing, metro-based routing, dissimilar
gateway protocol, etc) and none have gained traction. Why? But anyway,
I would certainly like to get RRG to have a broad view of ICN routing
efforts in general. Both groups could benefit tremendously.

> 2.- SDN-to-SDN routing and protocols, advanced peering between SDN domains beyond plain BGP, i.e., beyond IP-centric reachibility towards generalized application/service driven exchanges.
>
> Related to the latter scenario there some are emergent ideas on software defined Internet exchange
> http://www.cs.umd.edu/~dml/papers/sdx-ons13.pdf

I don't think of SDX as multi-domain SDN routing - rather, it seems to
me that it uses SDN as a tool for better routing of IP. Have you seen
the multi-domain SDN routing ideas developed in the OGF? The NSI
protocol is ugly and currently limited, but they (theoretically,
potentially) share state about whole networks, with intelligent
handling of multiple interconnect points, while allowing abstraction
and protection of private information. I've also seen a pretty good
effort to use lessons learned from MPLS PCE for multi-domain SDN.
Anyway, multi-domain SDN is a good topic.

Scott