[openstack-dev] [OpenStack] Cinder PTL Candidacy
john.griffith at solidfire.com
Sat Sep 21 17:59:18 UTC 2013
I would like to run for election as Cinder PTL for the upcoming I release.
For those who don't know me, I am the current Cinder PTL and have been
working on OpenStack for almost two years now. Most of that time has been
spent leading the Cinder efforts starting with the separation of the
Nova-Volume code. The upcoming Havana release will mark our third release
with Cinder as a released project in OpenStack and I think we have
continued to grow and get better with each release, Havana being no
Some of you may know that there's a lot that goes in to being a PTL. I see
it as a combination of Project/Team management, Technical Leadership and
Evangelism. I love the job, every aspect of it as well as the challenges
that come with it. This includes talking about Cinder, brainstorming new
ideas and *recruiting* new vendors and contributors.
Given Cinders unique plugin model (we now have over 17 supported back end
device drivers), one of the biggest challenges is continuing to provide and
maintain a feature equivalent and robust reference implementation as well
as maintaining consistent behaviors across all of these drivers. At the
same time we've provided ways for different vendors to expose their own
unique features via optimizations or the use of types and extra-specs. The
mantra here has been, if vendor A wants to implement a new feature there
has to be a way to implement it across the board. It may not be the most
elegant or efficient for some back-ends, but the idea is you will have
consistent capabilities and behaviors. One of the things that I think
makes the above described philosophy so powerful is that it drives interest
and contributions to the project as a whole. The result is that Cinder has
multiple contributors from competing storage vendors all driving to advance
the overall project for EVERYBODY. I think we've done a fantastic job in
this respect and if elected I plan to continue to drive this as one of the
main ideologies for the project.
There are a number of things that I see as needing particular focus during
the upcoming release, and if elected I will try and drive these items:
1. Functional test qualification for back end devices
Many of you have heard about this via mailing list, but I think it's
critical that we have some way to share publicly that the drivers that are
shipped with cinder actually work. This is something that I've started and
plan to implement for the I release, it will start simply by requiring that
all of the tests we currently run in the CI gates are run against each
driver/back-end and the results of those runs are submitted publicly.
2. More involvement in Horizon
This has been something that I've thought of and mentioned in the past but
it hasn't really happened. For the upcoming release I would like to drive
folks that implement new features in Cinder to help contribute and get
those same features exposed in Horizon. In my opinion OpenStack is growing
so quickly and becoming so large, that there needs to be more contribution
from all projects to keep Horizon updated. I'm not saying something as
extreme as to have a rule that a BP is not closed/implemented until the
Horizon work is done, but I'd love to see folks have that sort of view
(un-enforced norm in terms of behaviors) with features they implement in
Cinder as we go forward.
3. Organization and strategy around common libraries for Block Storage
This is something that we talked about during the last summit and we made
some progress on. The problem is that it started to grow in to it's own
beast and I think we lost our focus. I'd like to really emphasize early in
the Icehouse release what our common goals and objectives are here and make
this a reality early on in the release cycle (preferably the first
4. Continued implementation of task-flows and states
We've made a good initial start here by adopting task-flows for our
volume-create process. I think there's a lot more that can be done to
improve upon what we have here, and also to spread that out to the other
5. More interaction with the other projects
The worst thing for any project in OpenStack in the coming release IMO is
going to be working in isolation. There are a large number of new
projects/programs being introduced, many of which that utilize bits and
pieces of all OpenStack projects. I think it's critical that we come up
with some ways to collaborate across projects and programs, not only for
better implementations in consumer projects, but also to make sure we
provide valuable features that might be needed. This particularly includes
projects like Trove, Ironic and Triple'O.
6. Continue to grow the contributing community
This is key IMO for any Open Source project, we need to continue to grow
the interest and contributions. Whether that be via new drivers/vendor
participation or new core features, I believe involvement and particularly
community growth is a critical measure of a projects health.
There are a number of other things on my list, but these are the main
driving philosophies that I'd propose in Cinder for the next release cycle.
If you have any questions for me, or there are specific concerns or
omissions from this list don't hesitate to send an email or grab me on IRC.
I'm happy to share my opinions and ideas and clarify anywhere that I can.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the OpenStack-dev