Globus Alliance views on BES
Hi, sorry for the delay but we've put together some documents summarizing our Globus Alliance view on what the important issues and constraints are for a successful Basic Execution Service standard. Please see the attached documents: 1. Approach -- a high-level survey of our concerns 2. Rendering -- more detailed thoughts about interface designs 3. Roadmap -- simple thoughts on a phased approach Later, I will send an additional note with links to Globus Toolkit 4.0 draft documents for our WS-GRAM job management component, and a brief overview of the service interface for those who want to think in WSDL and not read a bunch of documents. This probably will not be ready for another 12-18 hours. Please feel free to respond on the list or privately with comments about agreement, disagreement, or relative priorities! I will try to incorporate any feedback I hear into planning for my ultra short presentation at the BoF in Seoul. Thanks, karl -- Karl Czajkowski karlcz@univa.com
Hi, Dave Berry was kind enough to point out that the terminology I used and referenced in the earlier BES position documents are not to be found in recent BES docs sent to the OGSA list by Steven. Sorry about that; I guess I've been juggling too many drafts of things floating around! Google to the rescue: http://www-unix.gridforum.org/mail_archive/drmaa-wg/2005/02/doc00000.doc In short: Actor : like a GRAM-launched job process Proxy Actor : like a GRAM job manager Container : the GRAM service that accepts jobs Annotations : meta-data, e.g. resource properties or external assertions? Application Instance Descriptor : job description, e.g. RSL Application Descriptor : something more abstract ;-) I think the spirit of that doc is good for this current discussion, in trying to sidestep some of these loaded terms such as "job". Not to say that the final rendered interfaces should persist in using such abstract names... karl -- Karl Czajkowski karlcz@univa.com
participants (1)
-
Karl Czajkowski