Open Stack

On 12/03/2014 11:04 AM, Maish Saidel-Keesing wrote:
> It is perhaps blasphemy - but Enterprise wants the same they can get 
> today with VMware - and the is standard HA.

I think everyone here knows that and also agree that VMware class of
products and OpenStack are as different as Microsoft Windows systems and
GNU/Linux ones. We're not going to make much progress if we keep
comparing apples and oranges.

There is a clear need identified by this working group: guests need to
be able to recover from failures and run persistent workloads. VMware
does this their way, OpenStack should pick whatever best way there is,
without assuming that VMWware has the only or necessarily the best solution.

On 12/03/2014 11:11 AM, Britten, Tyler wrote:
> Thanks for the references, really good stuff. I'm not sure what the 
> right answer is at this point, but I feel like it's definitely far 
> from a solved problem and at the same time something very important
> to enterprise adoption.

Russell's post has content, worth reading again. In there he gives
plenty of areas worth exploring further that *don't* extra require
development time in Nova. This is crucial because the objectives for
Nova (and all other projects, too) have already been set for the next 6
months [1] and most of them are *not* about new features.

So if this group wants to see progress in guest HA in this cycle I'd
seriously spend time investigating the suggestions in Russell's post.
They may turn out to be impractical but at least we'd be discussing on
solid technical grounds. Best case scenario, that pacemaker approach is
so simple that persistent workloads could be improved with better
documentation around this specific use case, with little code and in
relatively short time.

Who's volunteering to dive deeper in Russell's post?

/stef


[1] https://etherpad.openstack.org/p/kilo-nova-priorities-tracking