SysMan Mall

SuperComputing '06 - Tampa Florida

Touted as a rapidly growing way of doing super computing, High Performance Computing Clusters are the mainstay as it appears from various trends and data available at hand. This all becomes interesting given clusters with  Terra FLOPs (TFLOPs) are being constructed in a few days time which was if not unthinkable atleast considered of a certain degree complexity in the past.  As these TFLOP grids become common, talks about how Peta FLOP grids can be created is happening.  The beauty of all this is that these are being created using commodity computing components like x4100, x4200 type of hardware. This is different from previous HPC servers which were SMP based systems like Sun's E25k that could be populated with 100+ processors in a single physical system. These SMP mammoths demonstrated linear scale performance growth as processors were added. Newer approach is to take a rack full of systems like x4100 or similar components and take a few of these racks to create an HPC cluster.

 System Manager from that point of view is a great tool to carry our rapid systems deployment, right from the time you wire a system into a rack. Once discovered it is possible to apply power control operation and deploy OS on such a system. It then can go on to provide rich hardware information about the hardware being managed and some good OS specific information.

Heard a lot about how various market forces are shaping in this segment and what are the next challenges. Also got an opportunity to speak in front of a geeky crowd.  I'm already excited about tomorrow and am all set to see other vendors at SC'06 happening here in Tampa Florida.

Posted at 08:33PM Nov 11, 2006 by adikhit in General  |  Comments[0]