Hello, Duncan, 

Good questions. Currently, the availability zone (AZ in short) terms are not 
applied to both Cinder and Nova together, but seperately. That is to say, the 
AZ for  Cinder can has no relationship to the AZ for Nova. 

Under OpenStack cascading scenario, we would like to make each cascaded 
OpenStack function as fault isolation AZ, therefore, the AZ meaning for Cinder 
and Nova would be kept same. Now it's done by configuration. And if a volume 
located in another AZ2 (cascaded OpenStack)  was attached to a VM located in 
AZ1, it'll be failed, and should not be allowed. 

It's good to add AZ enforcement check in the source code of proxy (no need to 
be done on the trunk source code) to make sure the volume and VM located in the 
same cascaded OpenStack.

That's great you are interested in deep diving before design summit. Please 
follow this thread for the venue and date-time.

Best Regards

Chaoyi Huang ( joehuang )

From: Duncan Thomas [duncan.tho...@gmail.com]
Sent: 02 October 2014 22:33
To: OpenStack Development Mailing List (not for usage questions)
Subject: Re: [openstack-dev] [all] [tc] Multi-clouds integration by     
OpenStack cascading

On 2 October 2014 14:30, joehuang <joehu...@huawei.com> wrote:

> In our PoC design principle, the cascaded OpenStack should work passively, 
> and has no kowledge "whether it is running under cascading senario or not to" 
> and "whether there is sibling OpenStack or not", to reduce interconnect 
> between cascaded OpenStacks as much as possible.
> And one level cascading is enough for foreseeable future.

The transparency is what worries me, e.g. at the moment I can attach
any volume to any vm (* depending on cinder AZ policy), which is going
to be broken in a cascaded scenario if the volume and vm are in
different leaves.

> PoC team planned to stay at Paris from Oct.29 to Nov.8, are you interested in 
> a f2f workshop for deep diving in the OpenStack cascading?

Definitely interested, yes please.

OpenStack-dev mailing list
OpenStack-dev mailing list

Reply via email to