[gsa-rg] Use case document updated and pre GGF14 plans

[Ramin Yahyapour]

Donal K. Fellows wrote:
> Ramin Yahyapour wrote:
> 
>> I do not see that service guarantees always need advance reservation.
>> As I wrote in my example, for HPC systems many use some kind of
>> scheduling
>> strategy like backfilling for which you have to provide a maximum length
>> of your job.  On these systems, the local scheduler can usually easily
>> provide
>> you before/at submission time with the information when your new job will
>> be finished at the latest. However, you still do not know when your job
>> will actually run. It can (and usually will) be much earlier than the
>> expected
>> deadline. Therefore, from a practical point of view many existing
>> scheduling systems
>> from HPC could more or less easily provide you with such a service
>> guarantee.
> 
> 
> Other kinds of guarantee are also possible once you move into the
> economic domain. For example, "I'll run your job by this time or your
> money back!" OK, that's not suitable for all applications, but for many
> people that would be an acceptable trade-off (and you can imagine
> brokering agents carrying out probabilistic studies of the likelihood of
> a particular offer being honoured instead of the penalty clause being
> exercised, and so on).
> 

Yes. These are valid and relevant examples.
I would see these examples as part of an economic/market-oriented
profile that I mentioned later in my previous post.
There can be many (arbitrary complex) agreement types which bound the
contracting parties to many things (like for instance, refunding or
penalty payments). A sophisticated Grid Scheduling Architecture should
be able to cope with the arrangement/negotiations of such agreements.
I see this as a very advanced example as the participating schedulers
have to identify what kind of agreements and policies they support.
Therefore, I would propose to keep the simpler profiles with
advance reservation and service guarantees in terms of known allocation
bounds seperate from the economic profiles which will have to
employ more complex cost negotiations, refunding, or other features.

We might later consider to discuss typical agreement types which a Grid
scheduler should support in the economic profile.


Ramin

-- 
Dr.-Ing. Ramin Yahyapour       | mailto:Ramin.Yahyapour at uni-dortmund.de
Computer Engineering Institute | phone: +49 (231) 755-2735
University of Dortmund         | mobil: +49 (179) 5261973
44221 Dortmund / Germany       | fax:   +49 (231) 755-3251