Re: [mpls] Reference Augmented Forwarding - MPLS RAF

[Robert Raszuk <[e-mail]>]

Hi Tony,


> Under your proposal, it seems to be required.
>

How did you reach that conclusion ? Is there anywhere in my proposal some
text that precludes use of Entropy Label as is today? I would make sure you
can make good enough hash instead of doing such mapping as actions.

Assuming that ECMP is one of our network actions, then we want some entropy
> label (or similar value) to be considered when forwarding so that we
> achieve multi-path traffic separation.
>

IMO network programming is not really aiming for such usecase(s).


> #2 - Why don't you build your ingress to egress transport in a such way
> that 16 ECMP happens based on transport label with zero actions ?
>
> That would be an LSP with no EL.  You get one traffic path, not ECMP.
>

This is not what I had in mind. Why no EL ? When you send packets to an LSP
you make sure without MNA or RAF that network will do a good ECMP on it.
Just like today when neither MNA nor RAF exists.

Kind regards,
R.