[openstack-dev] [nova][tc] Seeking feedback on the OpenStack cloud vision

Zane Bitter zbitter at redhat.com
Wed Oct 24 15:06:34 UTC 2018

Greetings, Nova team!
As you may be aware, I've been working with other folks in the community 
on documenting a vision for OpenStack clouds (formerly known as the 
'Technical Vision') - essentially to interpret the mission statement in 
long-form, in a way that we can use to actually help guide decisions. 
You can read the latest draft here: https://review.openstack.org/592205

We're trying to get feedback from as many people as possible - in many 
ways the value is in the process of coming together to figure out what 
we're trying to achieve as a community with OpenStack and how we can 
work together to build it. The document is there to help us remember 
what we decided so we don't have to do it all again over and over.

The vision is structured with two sections that apply broadly to every 
project in OpenStack - describing the principles that we believe are 
essential to every cloud, and the ones that make OpenStack different 
from some other clouds. The third section is a list of design goals that 
we want OpenStack as a whole to be able to meet - ideally each project 
would be contributing toward one or more of these design goals.

The 'Basic Physical Data Center Management' goal was written to 
acknowledge Nova's central role in OpenStack, and emphasize that 
OpenStack differs from projects like Kubernetes in that we don't expect 
something else to manage the physical data center for us; we expect 
OpenStack to be the thing that does that for other projects. Obviously 
Nova is also covered by the 'Hardware Virtualisation' design goal.

The last paragraph of the 'Plays Well With Others' design goal was 
prompted by discussions in Cinder. I don't think the topic of other 
systems using parts of Nova standalone has ever really come up, but if 
it did this might be somewhere to look for guidance. (Note that it's 
phrased as completely optional.)

A couple of the other sections are also (I think) worthy of close 
attention. The principles in the 'Application Control' section of the 
cloud pillars remain important. Nova is a bit unusual in that there are 
a number of auxiliary services that provide functionality here (I'm 
thinking of e.g. Masakari) - which is good, but it means more things to 
think about. Not only whether any given functionality is needed, but 
whether it is best provided by Nova or some other project, and if the 
latter how Nova can provide affordances for that project to integrate 
with it.

The Partitioning section was suggested by Jay. It highlights the known 
mismatch between the concept of Availability Zones as borrowed from 
other clouds and the way operators use OpenStack, and offers a long-term 
design direction without being prescriptive.

If you would like me or another TC member to join one of your team IRC 
meetings to discuss further what the vision means for your team, please 
reply to this thread to set it up. You are also welcome to bring up any 
questions in the TC IRC channel, #openstack-tc - there's more of us 
around during Office Hours 
(https://governance.openstack.org/tc/#office-hours), but you can talk to 
us at any time.

Feedback can also happen either in this thread or on the review 

If the team is generally happy with the vision as it is and doesn't have 
any specific feedback, that's cool but I'd like to request that at least 
the PTL leave a vote on the review. It's important to know whether we 
are actually developing a consensus in the community or just talking to 
ourselves :)

many thanks,

More information about the OpenStack-dev mailing list