On Sun, Apr 25, 2010 at 10:40 PM, Russ White <[email protected]> wrote:
> I'm not certain it has anything to do with the technology. Rather, it's
> a matter of reality: if you want steer traffic, you must have more
> state. Either that state must be in the packet header (source routing),
> in the construction of a tunnel (semi-permanent circuit), or in the
> forwarding table. You can't get better traffic flow control with less
> information; it's just a logical impossibility.
>
Russ,

what are the underlying reasons to steer traffic?

I think one reason is to utilize the network as much as possible, make
use of all links that are available and share the load over them. Is
this a task the only the network should do or can the endpoints
participate in the traffic engineering as well?

If the routing architecture announces attachment points (aggregated in
DFZ)  to Internet - the local network is not advertised throughout the
DFZ - and having MPTCP on the endpoints the traffic engineering is
taken care by the endpoints. If congestion or failure occurs on either
attachment point the transport protocol shifts more traffic to the
other path.
Also, I find Valiant load-balancing interesting, but more studies
around that topic is needed.

So I think you necessarily don't need more state in the network if the
endpoints can take greater responsibility for their sessions and the
goal is to utilize all possible paths (attachment points). TE due to
other reasons need other tools.

-- patte
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg

Reply via email to