Re: [Bier] Deborah Brungard's Block on charter-ietf-bier-01-06: (with BLOCK)

[Lou Berger <[e-mail]>]

Alia,


On 2/23/2018 12:22 PM, Alia Atlas wrote:
> Ice & Toerless,
>
> I do understand the desire to keep the OSPF & IS-IS extensions in BIER,
WRT isis and ospf extensions - in the past we worked hard to ensure that 
we weren't fragmenting the industry by developing overlapping solutions 
in different WGs.  Generally this meant doing the protocol work in OSPF 
and ISIS - we moved the TE extensions to TEAS(actually its predecessor) 
as the protocol WGs basically saw their job as transport and we had so 
many different technologies with overlapping needs that they didn't 
really want to spend their time working the details to developing  
unified / single mechanisms. (For ISIS, we basically did the definition 
in TEAS and ran the doc process in ISIS.)

I think the principle of having one place own the extensions to a 
protocol to ensure there are no overlapping extensions are defined still 
makes sense.  Perhaps rather than looking to TEAS, we should leverage 
the formation of LSR and say that *all* OSPF and ISIS extensions must go 
through that group.  Certainly other groups like BIER will identify 
required extensions and TEAS will continue to be the central clearing 
house for TE architecture and identification of switching technology 
generic extensions -- but LSR could run the process on all extensions 
and ensure we're not defining different mechanisms for distribution of 
the same information.

Lou

> but
> if you think this is really
> traffic-engineering, then the architecture and signaling fall under TEAS.
> That WG can also do the
> signaling work as easily if necessaray.
>
> I'd strongly prefer not to delay the recharter for this aspect.
>
> I know that the draft talks about this as directed forwarding, but from
> what I read, I do see it as the
> forwarding plane for BIER-TE.
>
> Regards,
> Alia
>
> On Fri, Feb 23, 2018 at 4:32 AM, IJsbrand Wijnands <[e-mail]> wrote:
>
>> Toerless,
>>
>> Wrt to the proposed "No control-plane work will be done in BIER-WG”
>>
>>
>> We probably should be a little more specific on what is meant with control
>> plane. With BIER-TE there are two different levels of it. There is the
>> Controller (PCE) control plane and (potentially) the underlay signalling.
>> It is completely valid to use the IGP’s to distribute the BIER-TE MPLS
>> Labels, and let the PCE calculate the paths. The current IGP drafts don’t
>> distribute labels for BIER-TE, so similar drafts would have to be written