[openstack-dev] [all][tc] Languages vs. Scope of "OpenStack"

Fox, Kevin M Kevin.Fox at pnnl.gov
Tue May 24 19:38:06 UTC 2016

Frankly, this is one of the major negatives we've felt from the Big Tent idea...

OpenStack use to be more of a product then it is now. When there were common problems to be solved, there was pressure applied to solve them in a way everyone (OpenStack Project, OpenStack Users, and Openstack Operators) would benefit from.

For recent example, there has been a lot of talk about reimplementing features from Barbican in Magnum, Keystone, etc, and not wanting to depend on Barbican. In the pre-tent days, we'd just fix Barbican to do the things we all need it to, and then start depending on it. Then, everyone could start depending on a solid secret store being there since everyone would deploy it because they would want at least one thing that depends on it. Say, Lbaas, or COE orchestration, and then adding more projects that depend on it would be easier for the operator. Instead I see a lot of of trying to implement a hack in each project to try and not depend on it, solving the problem for one project but for no one else.

Its a vicious chicken and the egg cycle we have created. Projects don't want to depend on things if its not commonly deployed. Operators don't want to deploy it if there's not a direct reason to, or something they do care about depending on it. So our projects are encouraged to do bad things now and I think we're all suffering for it.

Cross project work became much harder after the big tent because there was less reason to play nice with each other. OpenStack projects are already fairly insular and this has made it worse. Opening up additional languages makes it yet harder to work on the common stuff. I'm not against picking 1 additional language for performance critical stuff, but it should be carefully considered, and should have to be carefully reasoned about the need for.


From: Ian Cordasco [sigmavirus24 at gmail.com]
Sent: Tuesday, May 24, 2016 12:12 PM
To: Jay Pipes; OpenStack Development Mailing List (not for usage questions)
Subject: Re: [openstack-dev] [all][tc] Languages vs. Scope of "OpenStack"

-----Original Message-----
From: Jay Pipes <jaypipes at gmail.com>
Reply: OpenStack Development Mailing List (not for usage questions) <openstack-dev at lists.openstack.org>
Date: May 24, 2016 at 11:35:42
To: openstack-dev at lists.openstack.org <openstack-dev at lists.openstack.org>
Subject:  Re: [openstack-dev] [all][tc] Languages vs. Scope of "OpenStack"

> On 05/24/2016 06:19 AM, Thierry Carrez wrote:
> > Chris Dent wrote:
> >> [...]
> >> I don't really know. I'm firmly in the camp that OpenStack needs to
> >> be smaller and more tightly focused if a unitary thing called OpenStack
> >> expects to be any good. So I'm curious about and interested in
> >> strategies for figuring out where the boundaries are.
> >>
> >> So that, of course, leads back to the original question: Is OpenStack
> >> supposed to be a unitary.
> >
> > As a data point, since I heard that question rhetorically asked quite a
> > few times over the past year... There is an old answer to that, since a
> > vote of the PPB (the ancestor of our TC) from June, 2011 which was never
> > overruled or changed afterwards:
> >
> > "OpenStack is a single product made of a lot of independent, but
> > cooperating, components."
> >
> > The log is an interesting read:
> > http://eavesdrop.openstack.org/meetings/openstack-meeting/2011/openstack-meeting.2011-06-28-20.06.log.html
> Hmm, blast from the past. I'm sad I didn't make it to that meeting.
> I would (now at least) have voted for #2: OpenStack is "a collection of
> independent projects that work together for some level of integration
> and releases".
> This is how I believe OpenStack should be seen, as I wrote on Twitter
> relatively recently:
> https://twitter.com/jaypipes/status/705794815338741761
> https://twitter.com/jaypipes/status/705795095262441472

I'm honestly in the same boat as Chris. And I've constantly heard both. I also frankly am not sure I agree with the idea that OpenStack is one product. I think more along the lines of the way DefCore specifies OpenStack Compute as a Product, etc. I feel like if every project contributed to the OpenStack product, we might have a better adoption rate and a better knowledge base for how to make new services scale from day 1. Instead, we are definitely a loose collection of projects that integrate on some levels and produce what various people might combine to create a cloud.

I'm also not entirely that the answer remains true with the different defcore programs. It seems like DefCore makes us define a minimum viable OpenStack {Compute,Object Storage} and then you can add to that. But those two things are "OpenStack" and everything else is a nice additional feature. There's nothing that makes Barbican or Magnum or Ceilometer a core part of OpenStack. Yet they're projects of varying popularity that different people choose whether or not to deploy. If OpenStack were a product, I'd think that not deploying Ceilometer would be the exception.

Ian Cordasco

OpenStack Development Mailing List (not for usage questions)
Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe

More information about the OpenStack-dev mailing list