Subject: Re: Update on ITR Pre-Proposal Resent-Date: Fri, 10 Dec 1999 08:22:22 -0500 Resent-From: Geoffrey Fox Resent-To: p_gcf@npac.syr.edu Date: Fri, 10 Dec 1999 01:40:40 -0500 From: Tomasz Haupt To: Richard Pritchard CC: bedford.1@osu.edu, root.1@osu.edu, mfw@brahma.ticam.utexas.edu, gcf@npac.syr.edu, kenf@osc.edu, parashar@caip.rutgers.edu, clint@brahma.ticam.utexas.edu, Connie Pritchard , welsh@superior.eng.ohio-state.edu, mpesz@brahma.ticam.utexas.edu, sadayappan.1@osu.edu, sbryant@brahma.ticam.utexas.edu, arbogast@brahma.ticam.utexas.edu, bromwich.1@osu.edu, jeaton@brahma.ticam.utexas.edu I have a several comments on the pre-proposal. 1. what are the actual deliverables? a) Web portal: seamless access to resources b) PSE (in a sense of CCM PSE: providing assess to knowledge for non-experts) c) visual authoring tools (composition of application from modules: dataflow, tightly coupled, etc) d) models integration, models interoperability e) algorithms (science science as oposed to computer science) f) functional (near to production quality) GASP There are obvious overlaps between the above. The current text glorify a) and concentrate on d) and e). I think b) and c) are of potential interest, too. 2. Is it assumed that Web portal is brought to this project as an asset? Or can we make a reserach topic of it? Definitely, it must be better defined (event though we probably cannot devote much space for it). What is the back-end ("grid", "virtual computer"). Where cycles comes form, what are authentication mechanisms, what are authorization policies, who pays for CPU, etc.. Who is the target audience? A scientist with a graphical UNIX workstation (and possibly palmtop, too) or the rest of us under mercy of Microsoft? The only reason to run a code is to analyze its results. How do we provide visualization capabilities, or visualization is out of the scope? A true Web portal should provide access to the desktop resources, too. Currently, WebFlow is a mixture of Java and CORBA. (btw, it represents a distributed component model, three-tier application server, not just merly ORB based system). Should we add COM/DCOM? Support for different capacities and capabilities of clients (hardware, software, humans)? From immersive environmet to a wireless palmtop? Handicapped users (XML is there!)? Collaboratory? Different component models, different event models, requirements for the grid interface, requirements for information services. Potentially there is a lot of opportunities for a non trivial compuer science research here. Are we going for that? Technicalities: HPCC communities typically assume that software is free, or home made. With the portal technologies we are entering a commercial world. What is our strategy? Avoid, duplicate, negotiate, pay? 3. PSE. It is not my area, and definitely many interesting issues there 4. items c) and d) are closely coupled. I believe that this is a very hard problem, a spectacular research topic. It should be discussed in the proposal, I think. Interoperability means metadata, protocols, discovery (data, resources, interfaces), data and objects interchange, events, and more. Shall we bring new techologies to the Portal? UML, XMI, XML, Jini, ... Automatization of wrapper generation for legacy applications that exposes possible interactions between (independently developed) models/simulations. What if they are working at many different spatial and temporal resolutions? Incompatible message-passing. Are we going to address (some of) these issues? They may have a critical impact on all elements of the project. 5) item e) I will leave for others. Anyway, it is already covered in the text. Richard Pritchard wrote: > All: > > I have attached the latest version of the ITR Pre-Proposal. Please note > that the first page is the "Summary" called for in the outline. The > remainder (which is supposed to be limited to 5 pages) is the main body of > the pre-proposal. I have used some figures given to me by Keith Bedford and > some I took from a CCM PSE briefing, plus I had some written inputs from > UT. Other than that, I tried to at least lay out a template and an outline. > As you can see, we are already over on page count, and most of you have not > even given inputs yet. I suggest that we NOT do each approach, background, > etc. by school, but that we simply say what we are going to do, how we are > going to do it, and how we will pull it off with five (or six) institutions > involved. (Geoffrey, I could use some guidance on what to say about > FSU/Syracuse.) > > Anyway, please take a look at this and start making suggestions for > improvement. I will be here at OSC through most of the day on Friday, then > I fly back to Austin. We will have printed copies to talk about at our > meeting on Monday. > > RHP > > ------------------------------------------------------------------------ > Name: ITRpreproposal2.doc > ITRpreproposal2.doc Type: Winword File (application/msword) > Encoding: base64 > Download Status: Not downloaded with message > > ------------------------------------------------------------------------ > > Dr. Richard H. (Dick) Pritchard > Director, Federal Programs > OSC (Ohio Supercomputer Center) > 1224 Kinnear Road > Columbus, OH 43212-1163 > Phone (614) 292-1163 Fax (614) 292-7168 > rhp@osc.edu > Now half-time at The University of Texas-Austin > Phone (512) 475-9290 > > "Never attribute to malice that which is adequately explained > by stupidity". -- Hanlon's Razor