BU's Share Compute Cluster (SCC)
summary
Boston University's Share Compute Cluster is a 20-rack, roughly 750‑host installation (about 16,000 cores across generations) hosted at the Massachusetts Green HPC Center with ~2,000 users from 80 departments. It runs CentOS 6 in production with a staged move to CentOS 7, schedules with Open Grid Scheduler, offers Singularity images, exposes a 4‑petabyte GPFS backend via NFS, and serves ~700 compiled applications. Typical utilization sits between 70–90%; largest jobs reach ~448 cores; monitoring uses Nagios and Ganglia and the system reliably supports campus research workloads.
AI-written summary · about
chapters (26)
- 0:00Introduction
- 0:28SCC Overview
- 1:50HPC
- 2:23Serial
- 3:01Parallel
- 3:55Top 500
- 4:40Trends
- 5:01No massively parallel system
- 6:24Who is this for
- 7:39Who are our clients
- 9:18Hosts
- 10:16Building Blocks
- 11:32Connectivity
- 12:20Power
- 13:09Cluster Version 7
- 14:13Open Grid Scheduler
- 15:42Batch Scheduler
- 16:43Storage
- 17:38Hardware
- 19:11Monitoring
- 21:38Environment Modules
- 24:09Utilization
- 25:02Operations
- 26:12Team
- 27:30Future
- 28:10Questions
