[c-nsp] ECMP v Link Aggregation ofr MPLS

[Mark Tinka]

On Friday, March 14, 2014 11:08:39 AM Lukas Tribus wrote:

> I would say one of the major drawbacks with LAG is that
> the links need to be connected to a single box on the
> other site, while with ECMP you can connect links to
> different boxes and as long as the cost are equal, you
> will load-balance.

Also, on some platforms, QoS needs to be applied on member 
links, and can't be applied on the LAG bundle itself. I try 
to stay away from such platforms if I can, or make sure this 
isn't an operational concern, e.g., placing it on core- 
rather then customer-facing ports.

As of now, with Cisco and Juniper, all major features are 
supported on LAG bundles, e.g., QoS, MPLS, IPv6, BFD, MPLS-
TE, data plane filters, e.t.c. Just make sure all the 
features you have on regular interfaces are supported on LAG 
bundles. This development is not always one-to-one.

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20140315/665b25bb/attachment.sig>