The rest of the morning was spend in some excellent discussions of data management issues, with presentations from MIT, Washington (Bill Yock), and Berkeley, with some lively discussion. The presentation slides are on the web site, so I won’t try to recreate them.
This afternoon kicks off the meeting portion of the week, with Tom Barton giving an update on grid technology. Tom asked Carlie Catlett, Bob Cowles, Ian Foster, Von Welch, and Christoph Witzig, Jill Gemmill, David Lifka, Jim Pepin, John Paul Robinson, and Renee Schuey, along with Ken Klingenstein what they’d like CIOs to know about grids.
What is a grid? (Jim Pepin) – Grid is an overloaded term. Grid today almost means cyberinfrastructure – different question than what’s a grid in the Globus Infrastructure sense.
Why bother with Grids? it’s a way of pooling resources, as both an economy of scale and because of how science gets done.
Charlie Catlett says “…the biggest problem is identity management”. Witzig says rolling out PKI credentials to everybody is not a viable option.
Teragrid are working on joining InCommon. Their initial plans are to use Shibboleth to the TeraGrid portal, to get TG credentials.
Bob Cowles – “Would you want to fly in an airplane designed using the Storm botnet?” – the resources contributed to the grid need to be high quality.
EGEE is probably the most successful of the grids – European e-science grid. They’ve made the most progress towards federated identity management.
Ken notes that grids are good for finding fungible computing resources, but not for real multitasking.
Sometimes the latency in networks is a real issue (Terry – “Physics Rules”).
Ken – the campus more than the national Grids deserve some play.
Jim Phelps – these things are commodities – they’re in a grid in that some of the people who use them might not be located on campus, but not in the classic sense of being largely available.