[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: GGF6 and draft JSIM doc
unfortunately I will not be able to participate in tomorrows discussions,
but here are my comments:
There's only one minor glitch in the UML diagram, where the link type
between ScheduledJob and JobData must be changed from association to
Maybe it would also make sense to add a section where changes to the
document are listed.
In order to facilitate the comparison between the LTC/NCSA model and this
proposal I've compiled the following list of classes in the models
Equivalent Classes (w/o associations)
Classes only in JSIM proposal
Classes only in LTC/NCSA model
My thoughts about this:
There's one system hosting the BatchService, which is maintaining the
BatchQueues and is scheduling the BatchJobs to run on some other systems.
If you exchange the class names according to the list above, the same is
true for the LTC/NCSA model.
In order to make such semantic information available to readers, it would
help to add some use case scenarios. We should reuse (and extend) the use
cases the NPi work group has developed so far.
I am not so sure whether we can get away without defining resources
(meaning groups of candidate execution hosts) to get the complete picture.
They need not to be fully fleshed out, but could exist as abstract classes.
Further, I like and support the approach to have a separate class for
All in all, the structures of both models are similar enough to give me
hope that we are on the rigth track. I wish you a productive session.
Mit freundlichen Grüßen / Kind Regards
Linux Technology Center
IBM Laboratory Böblingen, Germany
<flon@ISI.EDU> To: firstname.lastname@example.org
Sent by: cc:
owner-cgs-wg@grid Subject: GGF6 and draft JSIM doc
Please review the document at this link:
Sorry for the very late distribution. We will discuss this document at
GGF6 on Thursday morning.