S. Digel, 11 July 2006
As Steve described at the closeout workshop for DC2, we won't have a DC2-like DC3, but in terms of thoroughly exercising the ISOC and science planning by the collaboration science working groups there's plenty of impetus [not to be confused with momentum] for defining new sky models and generating serious volumes of simulated data.
Steve pointed out (in slide 8 of his presentation, copied below) some possible scheduling opportunities for 'service challenges'. The meaning of the term is not completely clear to me. It certainly means the challenge of generating and managing the data for the ISOC, and we can choose to make it mean milestones for science tool development.
Richard estimates that a generating a year's worth of data like we did for the 55 days of DC2 would take about 200,000 CPU hours and require about 30 Tbyte of disk space. This is after the sky model is defined. The CPU time would take ~4 weeks (or more) to obtain, and disk space may be an issue, so we certainly won't be surprised by a suddenly-appearing successor to the DC2 data set. In terms of getting real user-type people to commit to running the tools, the availability of a service challenge-type data set together with real analysis goals from the science working groups would help a lot, but in principle we could have an adequate Checkout 4 with just the DC2 data or gtobssim-generated data.
We'll hear at the science tools meetings about drivers for a 4th science tools checkout. I think that primarily they relate to shaking down the tools, with the advances being incremental and possibly mostly in utilities.