Re: vSphere and RHEV-M Design Questions

[Michal Fojtik <mf...@redhat.com>]

On 07/28/2013 11:43 PM, Sean Peterson wrote:

Hi Sean,

Apologize for the long delay.

> The vSphere driver currently enumerates machines be scanning through the
> observable data stores. Each datastore volume is considered to be a
> separate realm. Why does the vSphere driver go with this instead of
> mirroring the RHEV-M driver in using datacenters as providers and clusters
> as realms?

In time we wrote the VSphere driver, the current approach seem the best 
for the use-case we try to solve in Aeolus project.

I agree that it could be much better when we can supply the UUID of the 
datacenter in the API_PROVIDER variable, as we do in RHEV-M. Also as a 
benefit we could gain some speed by not search all available datastores.

Can you please fill a JIRA for this and mark it as a 'feature request'?

> For the RHEV-M driver, why does the storage_volumes method return the
> storage domains instead of the individual disks? Would it be possible to
> extend the RHEV-M driver to list individual disks per instance?

It is as long as the rbovirt gem we use for communication with RHEV-M 
supports it. I will need to look at this.

Cheers,
   Michal


-- 

Michal Fojtik <mf...@redhat.com>
Deltacloud API, CloudForms