[occi-wg] Using OCCI with infrastructure frameworks and local batch adapters [Re: Sessions at OGF 34]

[Andre Merzky]

Hi Alan, OCCI-WG,

what would you currently consider the best platform to play with OCCI,
for providing resources to ran real workloads against (for scalability
tests)?

Eucalyptus would in fact be great, as that is already deployed for us
- but that seems not an option.  Some queuing system frontend would be
great, as those are also available in our settings, but I understand
that this is somewhat off target for OCCI.

So, what backend could we (easily) deploy in order to use OCCI in a
experimental setting?

Many thanks, Andre.


On Mon, Feb 27, 2012 at 4:51 PM, Alan Sill <Alan.Sill at ttu.edu> wrote:
> Hi Gregor,
>
> I surveyed a few of the group members and leaders on these topics, and the
> overall sense seems to be that while the interfaces to framework projects
> such as OpenStack, OpenNebula etc. do exist and and maturing, there are not
> yet any projects underway to make direct connections from OCCI to local
> batch resource managers such as the ones that you mention.
>
> In other words, there is not yet (and not immediately on the horizon) a
> project to introduce the OCCI equivalent of GRAM for batch job submission.
>
> That is not to say that such an interface is impossible, but it might
> require the same sort of care that went into development of LRM adapters for
> GRAM, and possibly is a design project.
>
> Questions b) - d) that you posed relate to some of the above active
> projects.  My understanding is that these topics will be discussed at this
> week's OCCI-WG teleconference, details of which are at the link below.
>
> https://forge.ogf.org/sf/wiki/do/viewPage/projects.occi-wg/wiki/TeleconferenceCalls
>
> In particular, B) and D) are active, but work across the board on Eucalyptus
> has been hard to motivate, as in most people's experience the current
> Eucalyptus framework has been buggy, hard to work with and not very
> scalable, so has been a low priority to get working.  (If we understand this
> to be a surrogate for working with EC2-style interfaces and therefore for
> working with Amazon, it might be possible to generate more enthusiasm, but
> the work so far has been on truly open interfaces.)
>
> The meetings are scheduled  Wednesdays 10am EDT, 4pm CET, 3pm BST, Thursday
> 12am Tokyo according to the group wiki page.  Can I suggest that you join
> them to pursue some or all of these topics with the group?
>
> Thanks!  And hope this helps,
>
> Alan
>
> P.S.: Group members, please feel free to supply any corrections...
>
> On Feb 25, 2012, at 2:40 AM, Gregor von Laszewski wrote:
>
>> In a discussion with Andre, he indicated that it would be nice to have
>> some OCCI services on FutureGrid. As I have not followed much availability
>> of such services here are a couple of questions that came up
>>
>> a.1) how can a user on install his own OCCI services that interface with
>> SLURM
>>
>> a.2) how can a user on install his own OCCI services that interface with
>> Moab
>>
>> b) how can a user install an OCCI based service if there is an openstck
>> installation available for him and he has an account on
>>
>> c) how can a user install an OCCI based service if there is a eucalyptus
>> installation available for him and he has an account on
>>
>> d) how can a user install an OCCI based service if there is a open nebula
>> installation available for him and he has an account on (I think Open nebula
>> may come with an OCCI service?)
>>
>> It would be great to get answers from the experts so we can post them on
>> FutureGrid
>>
>> By the way the JSON object rendering seems a very good idea.
>>
>> Thanks
>>
>> Gregor
>>
>> <ATT00001.txt>
>
>
> Alan Sill, Ph.D
> Senior Scientist, High Performance Computing Center
> Adjunct Professor of Physics
> TTU
>
> ====================================================================
> :  Alan Sill, Texas Tech University  Office: Drane 162, MS 4-1167  :
> :  e-mail: Alan.Sill at ttu.edu   ph. 806-834-5940  fax 806-834-4358  :
> ====================================================================
>
>
>
>
>
> _______________________________________________
> occi-wg mailing list
> occi-wg at ogf.org
> https://www.ogf.org/mailman/listinfo/occi-wg



-- 
Nothing is ever easy...