Re: [spring] Managing "global" SIDs

[Jeff Tantsura <[e-mail]>]

some disagree:

"Hi all,

I think each vendor will uses its own NMS. So, it can’t be easy to manage global SID allocation with multi-vendor environment alongside multi-NMS. Therefore, PCE (standard) is better choice.

Regards,
Vahid
"

Cheers,
Jeff
On Jul 24, 2019, 4:35 PM -0400, Robert Raszuk <[e-mail]>, wrote:
>
> > Sure, my point was that you won’t need a “NMS per vendor”
>
> I don't know any network which would use "NMS per vendor" today. In fact the entire reason to develop your own NMS is to have your own interface to the network provisioning to be vendor agnostic.
>
> Best,
> R.
>
> > On Wed, Jul 24, 2019 at 10:29 PM Jeff Tantsura <[e-mail]> wrote:
> > > Robert,
> > >
> > > Sure, my point was that you won’t need a “NMS per vendor” and hence a need to agree on control plane protocol (PCEP). Abusing control plane for configuration… been there :)
> > > Specifically to PCEP point - PCEP creates ephemeral state, (not persistent across reboots), and hence rather unsuitable for configuration.
> > >
> > > Cheers,
> > > Jeff
> > > On Jul 24, 2019, 4:21 PM -0400, Robert Raszuk <[e-mail]>, wrote:
> > > > Jeff,
> > > >
> > > > > I think (sincerely hope) you are wrong, there’s a reason we have spent last 7 or so years working on YANG.
> > > >
> > > > Even if in the perfect universe all devices would support Yang models uniformly - customers are still going to use their private NMS systems except instead of CLI or xml as NNI to network elements they will use YANG models.
> > > >
> > > > r.
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >