Open Stack

lebre.adrien at free.fr wrote:
>> De: "Thierry Carrez" <thierry at openstack.org>
>> On the downstream side, a lot of the working groups would likely
>> benefit
>> from being turned into SIGs. As long as their focus is not purely
>> operational, and they are willing to go beyond expressing gaps and
>> requirements and tackle implementation, the SIG format is a lot more
>> adapted and would facilitate gathering development resources and
>> coordinating efforts. The most obvious targets are the use-case
>> oriented
>> groups: Scientific WG, Telco/NFV WG, Massively-distributed/Edge
>> clouds
>> WG, Public Clouds WG, Large Deployments WG. LCOO could also be turned
>> into a SIG, given that they are already tackling implementation and
>> pooling resources -- although they would likely need to have a more
>> defined scope and/or merge with the Large deployments WG for clarity.
> 
> After spending a few cycles through these different WGs (in addition to the FEMDC that I'm co-chairing, I have been trying to follow the actions performed in the Telco/NFV WG and Large deployment Team), I'm not sure whether each of them should be turned into a specific SIG. 
> The main issue we are all facing is the number of contributors who can allocate sufficient amount of time to make things progressing. Because of such issue, there is for instance an on-going discussion between LCOO folks and Curtis (chair of the Telco/NFV in CC) to integrate NFV/Telco discussions under the umbrella of the LCOO WG (If I'm correct this is the second try for the Telco/NFV WG and despite the relevance of such a WG, contributors still did not come :().
> 
> More generally, I think there is a lot of overlapping challenges between WGs. According to what I understood from the SIG proposal, it would probably make sense to try to identify those overlapping challenges and create SIGs accordingly (NFV use-cases involved generally several sites/DCs which leads to the need of operating distributed cloud infrastructures, Deploying/operating a large cloud systems require mechanisms/features that scale well, those mechanisms can be suited also for FEMDC infrastructures and reciprocally...).
I totally agree that a transition to SIGs gives us the opportunity to
refactor those groups to avoid overlap and fragmentation, and we should
take it. I like that the LCOO is focused on associating development
resources to specific goals (making it very much SIG-like in its
composition and resulting efficiency) but in terms of problem space it
is trying to address, it seems to overlap both with the Telco/NFV group
and the Large Deployments group, creating fragmentation in the resources
associated to each.

Personally I think we should have one SIG around Telco/NFV issues
(ideally with a LCOO-like approach to pooling development resources to
actually implement its priorities), one SIG around Large
Deployments/Scaling, one SIG around Edge/distributed clouds, and one SIG
around Public clouds -- because those all represent slightly different
problem spaces, and different actors that could be interested in pooling
their dev resources.

-- 
Thierry Carrez (ttx)