Already read of course - in fact the sequence support, from my perspective, is different from workflows as follows: The difference between Grid workflows and sequences also applies to compilation as well as to the described AMBER part. I had some longer discussions with Andreas about it and the borderline between JSDL and not-JSDL in this context is tough somehow. We basically agreed that workflows are not part of JSDL and shouldn't be - and thus the name sequence seems to be more appropriate, because it's somewhat different and more constrained: Of course it makes sense that compilation and execution are performed in one sandbox, otherwise the application might be compiled and installed by one user and then re-used by multiple other users. There is no exact boundary and you can realize it with both workflows and sequences, but when using the proposed sequences you can often circumvent data-transfers into the job sandbox and necessary configuration activities specifying locations of the source-code and such like. In addition, the core usage is often "1" while the whole major job might require "64000" cores just to provide some numbers. Often the codes are specific for some types of architectures, which is in terms of many-core even more and more evolving. Hence, both workflow/sequence steps are bound to one particular site - so why re-schedule a workflow step based on another JSDL?! We experienced that more and more end-users require compilation prior to production runs, at least in our environments in HPC. I understand this is different in HTC environments. The same can be applied to this AMBER sequence executions, where are a few very short running jobs need to be run before/after the main executable itself - just to set the data formats right (one example). The view from my perspective of course. ...oups...time to begin. Some thoughts, Morris
-- -----Ursprüngliche Nachricht----- -- Von: David Wallom [mailto:david.wallom@oerc.ox.ac.uk] -- Gesendet: Donnerstag, 5. August 2010 15:19 -- An: Riedel, Morris; pgi-wg@ogf.org -- Betreff: Re: AW: [Pgi-wg] DEISA/PRACE and EGEE/EGI Interoperabiltiy Use case Uploaded -- -- I have done this now I believe -- -- David -- -- -- On 05/08/2010 13:12, "Morris Riedel" <m.riedel@fz-juelich.de> wrote: -- -- > Thanks for this David! -- > -- >> -- -----Ursprüngliche Nachricht----- -- >> -- Von: pgi-wg-bounces@ogf.org [mailto:pgi-wg-bounces@ogf.org] Im Auftrag von -- >> David Wallom -- >> -- Gesendet: Donnerstag, 5. August 2010 13:34 -- >> -- An: pgi-wg@ogf.org -- >> -- Betreff: Re: [Pgi-wg] DEISA/PRACE and EGEE/EGI Interoperabiltiy Use case -- >> Uploaded -- >> -- -- >> -- Hi Morris, -- >> -- -- >> -- I am going to start a review on your use case within gridforge. I have -- >> most -- >> -- comments already thought out so shouldn't take too long. -- >> -- -- >> -- David -- >> -- -- >> -- -- >> -- On 05/08/2010 11:24, "Morris Riedel" <m.riedel@fz-juelich.de> wrote: -- >> -- -- >> -- > Hi PGI team, -- >> -- > -- >> -- >> -- TBD: DEISA/UNICORE scientific use case (Morris) -- >> -- > -- >> -- > sorry for being a little bit late with this, but I tried to be as -- >> precise as -- >> -- > possible with this use case. -- >> -- > -- >> -- > It is available in GridForge: -- >> -- > -- >> -- http://forge.gridforum.org/sf/docman/do/downloadDocument/projects.pgi-wg/docm>> -- a -- >> -- > n.root.input_documents.use_cases/doc16039 -- >> -- > -- >> -- > It would be nice if 1-2 could comment on this one. Thanks. -- >> -- > -- >> -- > Take care, -- >> -- > Morris -- >> -- > ------------------------------------------------------------ -- >> -- > Morris Riedel -- >> -- > Deputy Division Leader -- >> -- > Distributed Systems and Grid Computing Division -- >> -- > -- >> -- > Jülich Supercomputing Centre (JSC) -- >> -- > Info: http://www.fz-juelich.de/jsc/JSCPeople/riedel -- >> -- > -- >> -- > -- >> -- > -- >> -- > _______________________________________________ -- >> -- > Pgi-wg mailing list -- >> -- > Pgi-wg@ogf.org -- >> -- > http://www.ogf.org/mailman/listinfo/pgi-wg -- >> -- -- >> -- -- -- >> -- =================================== -- >> -- Dr David Wallom -- >> -- Technical Manager -- >> -- Oxford e-Research Centre -- >> -- University of Oxford -- >> -- Rm 160, 7 Keble Road -- >> -- Oxford -- >> -- OX1 3QG -- >> -- -- >> -- +44(0)1865 610601 -- >> -- =================================== -- >> -- -- >> -- -- >> -- _______________________________________________ -- >> -- Pgi-wg mailing list -- >> -- Pgi-wg@ogf.org -- >> -- http://www.ogf.org/mailman/listinfo/pgi-wg -- -- -- -- =================================== -- Dr David Wallom -- Technical Manager -- Oxford e-Research Centre -- University of Oxford -- Rm 160, 7 Keble Road -- Oxford -- OX1 3QG -- -- +44(0)1865 610601 -- =================================== --