This year will see major new systems coming online at unprecedented scale and performance. Due to personal involvement over the years, I am especially excited to see Sequoia come online. It’s a 20 petaFLOPS (peak) system based on IBM BlueGene/Q technology at Lawrence Livermore National Laboratory (LLNL). At Oak Ridge National Laboratory (ORNL), they are upgrading Jaguar with GPGPUs to 9x its current performance to the same 20 petaFLOPS (peak). And Blue Waters at the University of Illinois, in the news due to the switch from IBM to Cray, will come online with a performance of over 11 petaFLOPS (peak).
Guess what core file system technology provides the foundation for these massive systems? In all three cases, it’s Lustre.