LSF: Scheduling and Managing Cluster Jobs
LSF allows non-expert users to run jobs on the thousands of computers in the cluster. You log in to the “head nodes” at odyssey.fas.harvard.edu, and submit LSF jobs with the bsub command. The LSF system puts your job into a queue, and when a computer processor (“compute node”) is available, LSF sends the job to that computer to run. Because compute nodes and head nodes see the same file storage, you get the same result as if you ran the program on the head node.