openstack team mailing list archive

[Michael Still <mikal@xxxxxxxxxxx>]

On Fri, Mar 15, 2013 at 11:27 AM, Michael Still <mikal@xxxxxxxxxxx> wrote:
> Hi. I'd like to run for the TC Spring 2013 election. I am a senior
> software engineer at Rackspace in their OpenStack group, and have
> worked in a variety of "cloud devops" roles for the last seven years.
> I think my operations experience gives me an interesting perspective
> into where OpenStack should be going in the next few years.

I was just asked what my thoughts on integrated projects are, so I
figured I should address that in public.

First off, the TC "has oversight on all technical matters" [1], so
while project incubation and integration is a topic that the TC
handles, it isn't the only thing I'd expect a well running TC to
handle. Examples of other things that the TC should be caring about
are:

- what languages we develop in
- what libraries we standardise on and if the existing standards are
still the best choices (for example, is sqlalchemy still the best way
of handling database abstraction?)
- how our CI infrastructure helps us run the project

To specifically address the incubation / integration question, I think
that OpenStack runs the risk of becoming distracted with too many
projects. We need to remember that the foundation is still quite
young, as is our community in general. I personally think that
OpenStack should be focussing on Infrastructure as a Service problems
for now, with the intention of expanding that scope later when we have
a solid stable base of infrastructure projects.

I can see why it is attractive for projects to want to become
incubated, especially because it means they get a bunch of (mainly CI)
infrastructure "for free", as well as a lot of mind share from the
association. We need to remember that our CI resources are limited
however, and we need to be careful not to overload that team.
Additionally, each project we add makes it hard for us to define who
is distributing "OpenStack". Do they have to ship every integrated
project? Even if its not relevant to the product they're trying to
build?

We live in a complicated world though. I think each of these decisions
should be made on its individual merits based on the facts that are
true at the time. I also think the TC is on the right track here at
the moment, so I don't think we need to wrap a bunch of super formal
policy around this area at the moment.

Hopefully that helps.

Cheers,
Michael

1: https://wiki.openstack.org/wiki/Governance/TechnicalCommittee