Previous | Session 26 | Next | Author Index | Block Schedule
D. Dossa, C. Matarazzo (LLNL), S. Marshall (SLAC), C. Smith, R. Lambert (CTIO, NOAO), M. Butler, C. Cribbs, R. Plante (NCSA), D. Sweeney, J. Kantor (LSST Corp)
The LSST Project will produce more than 25GB per minute of raw astronomical images, not including auxiliary information needed to fully characterize and calibrate the data. One of the important science drivers for the Project is the identification and announcement of transient astronomical events such as supernovae. In order to generate such transient alerts quickly, generally in less than 30 seconds after the images are taken, a high-throughput computing infrastructure is needed with low latency between it and the instrument readout buffers. At the same time, the computing infrastructure must be capable of storing and re-analyzing petabytes of data at the LSST Archive Center(s) far from the mountain site. These requirements necessitate a tradeoff between computing and storage resources versus network bandwidths, to optimize overall system performance, reliability, and cost. We present our preliminary design of the LSST data management infrastructure, including the computing, storage, and networking infrastructure necessary to fulfill the requirements of the LSST Project.
Previous | Session 26 | Next
Bulletin of the American Astronomical Society, 37 #4
© 2005. The American Astronomical Soceity.