The MPLS WG Archive

[David Charlap <David.Charlap@marconi.com>]

Richard Spencer wrote many good comments which I won't duplicate in my 
reply.  Following are my comments.

Erblichs wrote:
> 
> 	I don't see how a OSPF pkt could recv a
> 	different path (when only 1 best LS path exists)
> 	other than the best path chosen for it by the
> 	SPF algorithm.

SPF may select more than one path.  That's the whole point of ECMP - you 
use the topology learned by routing and select all of the shortest 
paths.  The forwarding system can choose to use this information to do 
load balancing.

> 	Same for IS-IS?

Yes.

> 	Same for RIP?

Yes, but for obvious reasons, the concept of a "path" (best or 
otherwise) isn't meaningful in a RIP network.

> 	Yes, LDP leans routes, however, the LDP keeps
> 	the same routes until explicitly torn down.

Read it again.  It installs labels onto FECs.  It learns FECs from an 
independant routing protocol.

> 	And i'm a little confused:
> 
> 	  * Are you saying that that a strict MPLS LSP
> 	  path can be duplicated with OSPF? 

Not with OSPF alone, but it can be done without MPLS.  That's the point 
of source routing.  (Admittedly, source routing is inefficient.  I don't 
think any forwarding chips today process this IP option.)

> 	  * Are you saying that the forwarding decisions
> 	  / advantages the 3rd bullet of MPLS in RFC 3031
> 	  is not true?
> 	"This can not be done with converntional forwarding,
> 	since the the identify of a packet's ingress router
> 	does not travel with the packet".

Please read my statements in context.  I was referring to this staement 
of yours:

>>>      With MPLS, it is possible to have numerous
>>>      paths to a dst where non-equal paths exist. This
>>>      would allow a set of equal dst pkts (FEC) to traverse
>>>      different paths based on different characteristics.
>>>      These characteristics could be different ISP
>>>      origination src addrs, bandwidth, delay, priority,
>>>      SLAs, etc.

Without MPLS, you can also have numerous paths to a destination with 
non-equal paths.  You use the same topology learned by routing (OSPF or 
ISIS - RIP doesn't learn any topologies) and run the SPF calculations on 
subsets of that topology, based on whatever packet classification 
filters you like.  This is the definition of CSPF.

As long as all of the routers have the same sets of filters/criteria 
configured, you'll get consistent routing, with different categories of 
packets going in different directions.

MPLS makes this job much easier (by eliminating the need for all your 
core routers to do all this clarrification), but it isn't the only way 
to get the job done.

-- David

_______________________________________________
mpls mailing list
mpls@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/mpls