Re: [mpls] Reference Augmented Forwarding - MPLS RAF

[Robert Raszuk <[e-mail]>]

Hi Aijun,

Hi, Robert:
>
> How to add additional data to accomplish the In-situ OAM similar function
> based on your proposal?
>

That is one of the reasons I delayed the writeup. I concluded however that
growing packets at each hop may be harmful from multiple angles. Yes yes I
know it is cool for probe packets, but as I mentioned on the list we have
real issues to sufficiently synchronize to sub ms (or less) clocks across
WANs hence I see this of rather limited practical use. Please observe that
the current model still works fine with postcard OAM in-situ mode.

However, said that if there is a burning need and we figure out the
clocking sync issue we could define some PSD container to carry such data.
But this would be a separate draft extending it. For now I am not keen on
adding it.

If not, what’s the difference to apply the traffic policy via the
> management from your proposal?
>

If you call arbitrary network action as traffic policy then there is no
difference. RFV just points to network action or forwarding behaviour which
needs to be executed on a given packet. I guess this is a terminology
thing. But irrespective of this mgmt channel is just one option how the
network instructions can be distributed. There are other options too.

Many thx,
R.