Purdue FAIR This proposal addresses a very interesting problem with a talented team. I only rate this proposal as FAIR as I believe the team lacks key distributed computing (grid) expertise. The proposed project is novel and very suitable for ITR program. I encourage resubmission to similar programs with an augmented team but same management and basic concept. The P.I. does bring both computer science and structural engineering expertise which is positive. In more detail: 1) The proposal is set to address computational infrastracture and it is on this I judge proposal. Thus I discount some of the important engineering features outside the proposed scope 2) It is not clear if this proposal is intended to put significant effort into crisis(hazard) management. If so, the team should be augmented in this area. 3) My basic problem is the (lack of) discussion of approach to distributed computing. Naively I would approach problem as a collection of "mini-grids" with an event (message) based middleware. Peer-to-peer network technology also appears relevant. The reason to think about these current hot areas (P2P, Grid) is to exploit the rich software infrastructure being developed for them. I understand the importance of real-time constraints but these will be needed for large scale grids. The team should address these points in the proposal and either show Grid/P2P expertise is not needed or include them in both proposal and PI expertise 4) One can see lack of deep discussion of architecture in discuission of "large structures". How large is large and what is implication for computational infrastructure? ============================================================================== Arizona FAIR I reviewed 4 ITR proposals in the Grid area broadly defined. This I would rate as the weakest proposal but with technically the strongest team. The major problem with this proposal is the very loose coupling between the science and the computer science. One can see this from the organization Chapter 3 is the science Chapter 4 reads like some sort of summary Chapter 5 is a collection of generally good computer science projects but they read as though one put together a collection of research projects solicited independently of the science Section 6 puts it all together but this is high level -- it leaves most of work in section 5 as excellent CS research uncoupled with supernovas Now this rather broad criticism is coupled with a personal predjudice -- namely I have some knowledge of the Grid and happen to think this is important technology but NOT particularly suitable for supernova studies where the advantages of distributed computing are obvious for "seamless access" but do not have the compelling case that can be made for applications exploiting what some term the expected data deluge -- high enery physics and observational astronomy can this way naturally exploit Grid technologies. Thus for me a close coupling between scientific requirements and CS projects is important as I am rather sceptical. There are some CS activities that can be made synergistic with the science -- the work of section of 5.1.1 is one but even here the coupling is not made explicit as though the CS team has yet to look at the applications (and vice versa) I would challenge the CS discussion when it claims (sec. 5.2) to build on CCA and Web Services. Many including me consider web services as obvious "wins" for loosely coupled problems and consider the CCA to be a brave but unproven attempt to apply web service ideas to the much harder closely coupled problems (as in Supernovas) where web services and Grids appear less applicable. It is "good" to improve on the CCA but this is so challenging that some hint of key ideas should be given -- otherwise it looks like wishful thinking. ========================================================================= Florida EXCELLENT This is a wonderful concept and a very well written proposal. This was clearly the best of the four ITR proposals that I reviewed. It has a well established team that already is collaborating. The computer science is very well integrated into all aspects of the project. I believe that the basic technical approach -- web services and event based middleware -- is the right idea.This proposal has my strongest support for funding I had a few not so important comments 1) I didn't care for the virtual machine discussion in section C 4.5 Virtual machines are good for the distributed enironment of AMIGA and more generally the Grid but the VM of VMware etc. are not very similar to those we will produce with Web Services 2) The proposal team could usefully study peer-to-peer networks I think they could be useful 3) I make my ranking of proposal even though the team involved is less well known in the CS community than those on other 3 I looked at. ========================================================================= Ohio State GOOD This is a well integrated and well written focussed proposal without however much innovation. I would consider this good incremental research by a quality team with an excellent track record. I expect them to succeed. A couple of comments 1) I would have preferred to see a greater emphasis on bio-informatics cyberinfrastructure -- Saltz is very well placed to lead the development of this. 2) I suggest the team look at the emerging web service approach to filters of the type like data-cutter. WSFL was an original approach and this area continues to change rapidly. You should not adopt "the fad of the day" but learn lessons and be ready to adopt technologies (like Web Services WSDL) as they emerge.