Open Stack

On Fri, 17 Aug 2018 16:51:10 +0100 (BST), Chris Dent wrote:
> 
> Earlier I posted a message about a planning etherpad for the
> extraction of placement
> 
>     http://lists.openstack.org/pipermail/openstack-dev/2018-August/133319.html
>     https://etherpad.openstack.org/p/placement-extract-stein
> 
> One of the goals of doing the planning and having the etherpad was
> to be able to get to the PTG with some of the issues resolved so
> that what little time we had at the PTG could be devoted to
> resolving any difficult technical details we uncovered in the lead
> up.
> 
> One of the questions that has come up on the etherpad is about how
> placement should be positioned, as a project, after the extraction.
> The options are:
> 
> * A repo within the compute project
> * Its own project, either:
>     * working towards being official and governed
>     * official and governed from the start
> 
> The etherpad has some discussion about this, but since that etherpad
> is primarily for listing out the technical concerns I thought it
> might be useful to bring the discussion out into a wider audience,
> in a medium more oriented towards discussion. As placement is a
> service targeted to serving the entire OpenStack community, talking
> about it widely seems warranted.
> 
> The outcome I'd like to see happen is the one that makes sure
> placement becomes useful to the most people and is worked on by the
> most people, as quickly as possible. If how it is arranged as a
> project will impact that, now is a good time to figure that out.
> 
> If you have thoughts about this, please share them in response.

Thanks for kicking off this discussion, Chris.

I'd like to see placement extracted as a repo within the compute 
project, as a start. My thinking is, placement was developed to solve 
several long-standing problems and limitations in Nova (including poor 
filter scheduler performance, parallel scheduling races, resource 
tracker issues, and shared storage accounting, just to name a few).

We've seen exciting progress in finally solving a lot of these issues as 
we've been developing placement. But, there is still a significant 
amount of important work to do in Nova that depends on placement. For 
example, we need to integrate nested resource providers into the virt 
drivers in Nova to leverage it for vGPUs and NUMA modeling. We need 
affinity modeling in placement to properly handle affinity with multiple 
cells. We need shared storage accounting to properly handle disk usage 
for deployments on shared storage.

As we've worked to develop placement and use it in Nova, we've found in 
most cases that we've had to develop the Nova side and the placement 
side together, at the same time, to make things work. This isn't really 
surprising, as with any brand new functionality, it's difficult to 
fulfill a use case completely without integrating things together and 
iterating until everything works. Given that, I'd rather see placement 
stay under compute so we can iterate quickly, as we still need to 
develop new features in placement and exercise them for the first time, 
in Nova. Once the major aforementioned efforts have been figured out and 
landed with close coordination, I think it would make more sense to look 
at placement being outside of the compute project.

Cheers,
-melanie