On May 5, 2017, at 11:52 AM, Stewart Bryant
<stewart(_dot_)bryant(_at_)gmail(_dot_)com> wrote:
On 04/05/2017 21:20, Alvaro Retana (aretana) wrote:
On 5/2/17, 12:57 PM, "Stewart Bryant"
<stewart(_dot_)bryant(_at_)gmail(_dot_)com> wrote:
Stewart:
Hi! How are you?
Thanks for the detailed review!
A significant part of the justification seems to evolve around the
inability of MPLS to function in an IPv6 only network.
It seems to me that this statement summarizes many of the concerns you
listed as Major in the review. I can see why it seems like the
justification is: “because MPLS doesn’t work, then we have to do IPv6.” –
but I think that even if a complete solution exists (for an MPLS deployment
on an IPv6-only network), some operators would still make the design choice
of preferring an IPv6-only deployment.
I think that it would be good for the authors to refocus the justification
away from “because X doesn’t work”. Would that address this part of your
concerns?
Thanks!
Alvaro.
Hi Alvaro,
That would be a step in the right direction particularly if Carlos' OPS
comments were also addressed.
Without going back over the detail, I seem to remember I had a question as to
how the homenet case worked given that SR needs topology info, and homenet
has chosen a DV protocol.
HNCP performs topology discovery and prefix assignment within the home network,
independent of the routing protocol being used.
- Mark
I also had a concerns about the validity of the scaling justification, and I
think there are issues of trust that need to be discussed.
Both of those may be addressable simply by providing more detail as Carlos
suggests.
Alternatively maybe it would be better to have a single use case: Operators
that wish to deploy SR without an MPLS control plane, although as I note, you
don't need an MPLS control plane to make MPLS SR work.
- Stewart