Re: [mpls] [sfc] Working Group adoption of draft-farrel-mpls-sfc

[Robert Raszuk <[e-mail]>]

>
> NP memory state is enormously more expensive ($$$) than route processor
> state.
>

​100% agree. ​

SR pushes that expensive state to the most cost sensitive components in the
> network (the ingress routers). The payback is a reduction in path state in
> the core.
>

​So to be practical let's face the question how many services will be
traversed ? 1 ? 2 ? 3 ? Is this so much to worry about as far as transport
state is concerned in ingress node ? ​Keep in mind that with current
technology ingress may actually (and very often will be) the host not the
router for SRH insertion (especially in v6 deployments).

​In draft-ietf-mpls-sfc you may "win" the less state on the ingress, but
you loose big time as now SF need to know where to forward given packet
further by properly setting the outermost transport label ​after N-th SF
processing. Clearly you need good control plane assist for it. IMO so much
for big savings ... Opex complexity increases exponentially when you start
to troubleshoot it.
​​

> If you have end to end NSH, then use it. Both of these drafts focus on
> what to do if you do not have end to end  NSH technology available.
>

​It would be very helpful to say so in this draft in bold :). ​

Anyhow worse things get standardized these days so consider this to be my
last note on that one ....

Cheers,
R.