[Openstack] Proposed OpenStack Service Requirements
abustardo at g2ix.com
Sat Feb 12 01:52:33 UTC 2011
On 2/10/11 7:35 AM, Jay Pipes wrote:
> On Wed, Feb 9, 2011 at 11:19 PM, Aimon Bustardo <abustardo at g2ix.com> wrote:
>> I have seen another thread on this also "Multi Clusters in a Region ".
>> the URI based naming makes the most sense here and the actual name is
>> each child is free-form but must conform to acceptable URI characters
>> for optional DNS. The hierarchy is location based only in as much as you
>> make it.. its a hierarchy that you can build with zones and clusters.
>> This can be in one physical location or many, but it will always be a
>> collection of servers and services. The hierarchical URI model fits
>> this perfectly and adds the benefit of being able to associate DNS with it.
> What I was saying is that zones don't mean "geographical regions".
> Zones are merely a collection of hosts or other zones, nothing more or
> less. There should be no connotation that a zone is equal to a
> geographical location. This allows Nova the most flexibility for
> organizations that want to set up zones that have nothing to do with
> geography, but rather want zones that represent hardware or service
> level agreements.
> Hope that explains which I'm uncertain of the benefits of a URI based
> naming scheme for zones *being enforced by Nova*. Nothing wrong with
> an organization naming their zones with a URI naming scheme. I'm just
> saying Nova should not *enforce* a URI naming scheme...
Ok, so you are saying : Yes on hierarchical, No on enforcement if URI
conventions. We are roughly on same page. However I caution not
enforcing URI based naming.. What if they use an automated DNS in
OpenStack? What if they add it later? Do they have to rename everything?
Aimon Bustardo | Morph Labs | +1.310.625.0608 (mobile) | +1.310.437.4898 (office) | www.morphlabs.com | aimon at mor.ph
More information about the Openstack