Parallel Workloads Archive

This page contains an archive of information regarding the workloads on parallel machines. It has two main parts: raw workload logs from various machines around the world, and workload models (sometimes derived from these logs). The goal is to make this information freely available to researchers interested in the evaluation of parallel systems, and specifically schedulers for such systems. In addition, there is a bibliographical listing of papers related to workload issues, and a list of people working in the field.

If you use the resources available here in your work, please cite this page as the source, so that whoever reads your work will also be able to obtain them.

This page is maintained by Dror Feitelson. Please send comments and additional information to .

• The standard workload format

• Workload logs

• Workload models

• Bibliography

• Workload modeling book draft

• People

• Related links:

  • The Grid Workloads Archive at TU Delft
  • MAUI scheduler traces from supercluster.org
  • NPACI JOBLOG Repository from SDSC

  • The Failute Trace Archive
  • The Tracefile Testbed with logs of MPI calls and other events in parallel applications
  • The BYU Performance Evaluation Laboratory Trace Collection Center, with address, instruction, and disk I/O traces.
  • New Mexico State Univ Trace Database -- address traces for processor architecture studies
  • AppLes survey regarding moldable jobs

  • Workshops on Job Scheduling Strategies for Parallel Processing
  • IISWC: IEEE Intl. Symp. on Workload Characterization, (previously called WWC: Workshop on Workload Characterization).
  • CAECW: Workshops on Computer Architecture Evaluation using Commercial Workloads

  • Cooperative Association for Internet Data Analysis (CAIDA)
  • DatCat Internet Measurement Data Catalog
  • Internet traffic archive
  • NLANR Internet traces