MareNostrum to utilize Moab and SLURM for enhanced performance

Moab Cluster Suite from Cluster Resources is an advanced workload manager and scheduler capable of optimizing scheduling and node allocation decisions. The software will allow BSC to gain extensive control over which jobs are considered eligible for scheduling, how the jobs are prioritized and where the jobs will run.

Moab also simplifies and unifies management and acts as a flexible policy engine that guarantees service levels and speeds job processing. Moab will be used as an external scheduler for the SLURM resource manager. SLURM, developed at Lawrence Livermore National Laboratory and Hewlett-Packard, is an open-source resource manager utilized on some of the most powerful supercomputers, such as BlueGene/L and ASC Purple.

The combination of Moab and SLURM will optimize the system performance of MareNostrum, allowing BSC to more efficiently conduct all scientific projects. Both tools will be deployed and installed by February.

“One of the important characteristics of the Moab-SLURM solution is its ability to handle applications simultaneously running across very large number of MareNostrum’s processors and using a large number of processors is key to BSC’s users,” said Sergi Girona, Operations Director of BSC. “The BSC’s Computer Sciences Department has also a lot of expectations with Moab because it allows dynamic integration and evaluation of new scheduling and resource allocation algorithms.”

MareNostrum’s adoption of Moab continues the trend of the largest and most powerful supercomputers in the world adopting Moab, as currently 5 of the top 6 systems on TOP500 have Moab. Leading supercomputing centers currently using Moab include the Sweden National Supercomputer Center, the National Center for Supercomputing Applications (NCSA), Lawrence Livermore National Laboratory, and CERN (European Organization for Nuclear Research).