Flux: a next generation resource manager for HPC and beyond

James Corbett
Lawrence Livermore National Laboratory

Flux is a next-generation resource manager that has been in development at LLNL for a decade. It is currently deployed on a handful of small clusters at LLNL, and is the planned resource manager for the El Capitan supercomputer. Flux offers numerous advantages over older resource managers like Slurm, LSF, and PBSPro. Among other things, Flux offers fully-featured scheduling; a standardized interface; feature-rich Python, Lua, and C libraries for submitting and monitoring jobs; and the ability to launch Flux within the resource allocations of other resource managers, or of Flux itself. The advantages of Flux and the ease with which it can be built and started have led many user-led successes at LLNL and other sites.

Presentation (PDF File)

Back to Long Programs