Open Stack

Hi Arkady, Thierry, Julia, 

On 30 November 2020 17:30:03 CET, "Kanevsky, Arkady" <Arkady.Kanevsky at dell.com> wrote:
>Dell Customer Communication - Confidential
>
[...] 
>In my view, interop cam at the right time when there were a lot of
>churn and a lot of implementations, and unclearness which openstack
>projects work together.
>We are past that stage for OpenStack.

I would argue that InterOp came too late.
The OpenStack ecosystem had already diverged significantly, when it started. 

So the choice was between setting a strict standard that would have achieved a very high level of interoperability but risking that most existing OpenStack implementations at the time would not actually meet the standards. Or setting the bar low, so getting most implementations in at the price of InterOp being insufficient to achieve interop for many use cases. The tradeoff was difficult - in the end the bar ended up pretty low but with a plan to increase it over time. That has actually happened, but not enough IMVHO. (And I accept blame for not pushing hard enough and not investing enough effort to increase more when I was working in the InterOp WG.) 

I really like that the InterOp program was mainly implemented as automated tests - RefStack with the guideline tests is something that I would expect every single OpenStack cloud to have included as a baseline test in their CI. 

On the board's role:

The board in the past was the guardian of the OpenStack brand - thus the approval of the guidelines there. But I agree with Julia - the technical depth required for the decision was not a good fit for a board. What the board would reasonably have done is to discuss the strategy behind the InterOp/Trademark program and leave the technical details to the InterOp group - which it mostly did in practice. 

Going forward, the board is the guardian of the OIF brand, no doubt, so I would assume that OpenInfra branding/certification/interop/compliance/... programs would still need oversight from the OIF board.

For the projects (OpenStack, Kata, StarlingX, Airship, ...) I think that delegating the responsibility for programs around the respective brand to the TC (or a similar body) of that project would make sense.

This does not preclude us from also having OpenInfra level programs that integrate several of our projects together. (We should then still delegate the technical details to technical experts and focus board discussions on the strategy of any such programs.)
Our main role here is to encourage collaboration, as was correctly said before and I perceive the OSF/OIF staff and board have a good track record here that we should continue. 

Just my 0.02€.

Looking forward to good discussions next week - thanks for bringing this up, Julia! 
-- 
Kurt Garloff <kurt at garloff.de>, Cologne, Germany
(Sent from Mobile with K9.)