Reviewer Feedback on Proposal

Software dependencies, networking, storage, and computer science innovation

· The proposal did not adequate incorporate a component of innovation beyond the non-trivial requirements of rapid scaling in multiple CI dimensions to serve the VO communities

· Rather, the proposal suggests that others will solve basic problems while OSG concentrates on integration, operation, and outreach.

· Joint software releases, [with EGEE, Teragrid, ...] a strong commitment to sharing computing jobs, and a robust accounting/settlement system are missing from this proposal

· So the emphasis must be on the intent to provide access to 10s of PBs of storage

· It is unclear if there is significant development [“OSG software stack”] work required. If so, the proposal should call out specific work required.

· Software releases (which consist of a virtual data toolkit that includes Condor, Globus, and NMI) seem to be dependent on the release schedule of the contributing components. The implications of this are not addressed

· There are no industry partners mentioned.

· The proposal makes no mention of commercial software or leveraging industry’s increasing interest in grid computing.

· Networking technologies ???

Effective Project Management

· The goals and milestones are rather high level to judge in detail

· I would like to have seen quantative metrics for success proposed for the future

· The challenge (which is common to other similar projects) is to management of to ensure cohesion of staff over a five year period

· The proposal provides little in the way of information about the operation of the OSG….. It is a long list of statements that have no support

· The panel is very concerned with the lack of specifics in the proposal and the highly distributed management structure. The panel suggests a more centralized management and a set of deliverables. The panel would like to see outside review of progress toward deliverables to make sure that the highly scattered work is advancing in coordination. The panel would like to see a timeline with responsible people and relationships between components. The panel was concerned that many of the PIs are overcommitted.

· There are no metrics that can be used to monitor progress in this proposal

· Who is ultimately in charge of this project? The PI? The executive director? It cannot be managed by committee

Connection to Science

· High-energy physics can certainly use the additional storage and opportunistic processing, but its availability will likely not be the basis for scientific breakthroughs.

· Most of the items identified in the proposal seem like straightforward implementation of capabilities that are already understood. What is the potential to make a major scientific advance in a specific domain

· One wonders how OSG will meet the massive computational needs of high-energy physicists, especially since the plan is to bring to physicists fewer CPUs that would be available thru two BlueGene/L computers.

· This reviewer cannot find evidence in the proposal that the proposed OSG extensions will lead to a major scientific advance in high-energy physics

Serving other than HEP/NP

· The applicants are largely from the NENP community and thus the other problem domains may bet less focus or understanding

· Given that the primary audience for this facility has been (and continues to be) physicists, it may take some effort to enable it and make it attractive to other communities

· It is stated that the OSG actively engages new research communities. Who have been brought into the fold beyond the physicists ?

Ease of use and adoption

· This would be eased by some form of portal development which makes access easier or more transparent