The MPLS WG Archive

[Curtis Villamizar <curtis@workhorse.fictitious.org>]

In message <3FB8F9A5.5010306@interflect.com>, mark seery writes:
> David Allan wrote:
> 
> >BTW, would be nice to get some other opinions, IMHO this is rather
> >fundamental....
> >
> One of the fundamental precepts of routing, is that each node 
> understands the routing decisions each other node makes, this is to 
> prevent black holing among other things. So for example, the SPF 
> algorithm is not a secret, precisely so that each router in a network 
> can make consistent decisions. As ECMP algorithms impact the forwarding 
> direction of packets, then I would suggest while not on the same level 
> of criticality as SPF, there is a similar logic that could be applied to 
> this case. I have similar thoughts about end-to-end QoS (but that's way 
> OT, just sited as another example).

That was never the case.

Each OSPF node did not have to know whether the downstream node did
ECMP and if it did whether it was a dumb per packet split or a hash
based, or something else.  It only had to know that the next node
downstream kept the packet going in the direction of lower OSPF cost.

> In addition, having worked at a vendor trying to work out the best way 
> to do ECMP (which is not easy task in an MPLS environment) it sure would 
> have been nice to have an industry-agreed way of doing this. While I 
> realise 802.1ad have gotten away with not agreeing to alogorithms, that 
> is on a point-to-point basis. When an algorithm impacts end-to-end 
> (within an area/AS at least) then I think the need to have an agreed 
> upon standard is more important.
> 
> So bottom line is I would have to agree that two routers from different 
> vendors can not communicate in the same network using ECMP unless the 
> algorithm is known by both. So if ECMP is in an architecture or 
> framework document, then it is not a proprietary feature, and therefore 
> it should be standardized.

There is plenty of existance proof that contradicts the assertion in
your last paragraph.

Curtis