[ogsa-rss-wg] First hack at slides for F2F

Soonwook Hwang hwang at grid.nii.ac.jp
Thu Aug 11 22:46:17 CDT 2005


The question (3) is interesting to me as well.
Mathias, what do you have in mind
dynamic information that EPS may want to retrieve
by interacting with the IS? Could the dynamic information 
not be retrieved by CSG from the IS for locating
candidate resources that fit the job's resource 
requirements? I am just wondering what kind of dynamic
information is there that EPS might need rather than 
CSG.

Soonwook  

-----Original Message-----
From: owner-ogsa-rss-wg at ggf.org [mailto:owner-ogsa-rss-wg at ggf.org] On Behalf
Of Donal K. Fellows
Sent: Thursday, August 11, 2005 10:46 PM
To: Mathias Dalheimer
Cc: ogsa-rss-wg at ggf.org
Subject: Re: [ogsa-rss-wg] First hack at slides for F2F

Mathias Dalheimer wrote:
> (1) In general, I agree with the architecture picture. But my 
> understanding of EMS was that the CSG should be interchangeable 
> independently of the EPS. This can be done in two ways: We could pass an 
> EPR to the EPS refering to the CSG, or the Jobmanager could handle the 
> interaction with both EPS and CSG. The latter one may be handy when 
> dealing with workflows. I remember Andrew Grimshaw arguing about it - 
> could you clarify this during the F2F.

I've not stated how the bits and pieces find out about each other.
Indeed, that's a part of the architecture that we've not touched on yet.

> (2) We should state that we are working with the GSA-RG to review their 
> use cases. I included a scentence in the version attached.

Thanks.

> (3) How do we deal with the information model people? IMHO, we have to 
> state what we expect them to include. For the CSG, things are easy: We 
> need to decide whether a JSDL job fits on the machine. The EPS uses 
> dynamic information which is more difficult, we need discussion about it.
> 
> BTW: In your architecture picture, the EPS is not interacting with the 
> information system. This should be the case.

Perhaps. I've instead been thinking about a model where the EPS does not
interact with the IS directly. This is going to be an interesting topic
of discussion I feel. :^)

> (4) I would prefer a neutral background, e.g. the GGF template.

I couldn't find the template. I've found something though (including the
IP statement) so that's incorporated now.

See the updated version attached.

Donal.





More information about the ogsa-rss-wg mailing list