Workload Characterization and Modeling Book – Data

This page contains links to the data sources used in the book Workload Modeling for Computer Systems Performance Evaluation.

Parallel Workloads Archive

• NASA Ames iPSC/860
• LANL CM-5
• SDSC Paragon
• CTC SP2
• KTH SP2
• SDSC SP2
• SDSC Blue Horizon
• HPC2N Cluster
• SDSC DataStar
• PIK IPLEX
• CEA Curie

Local Data from HUJI

• HU-CS user files dump from 2005: undergrads, masters students, phd students, staff (creation, modification, size)
• cspar file sizes 2003 (sizes only)
• inferno sessions 2004: dialup, students, and staff (sanitized 'last' output, 3 + 2 weeks)
• inferno processes 2003: ps bug (sanitized 'lastcomm' output)
• pita processes 1998: robot activity by user 13 and root (sanitized 'lastcomm' output)
  (or version with improved time resolution)
• HU-CS processes 2005 (sanitized 'lastcomm' output).
• Word frequencies in the book as of 5 March 2014 (rank, count, word).

External Data

• Unix file sizes survey – Gordon Irlam (no longer available, but cached copy of original and supplementary data posted here with permission)
• Unix process runtimes – Mor Harchol-Balter and Allen Downey
• BYU memory access traces (no longer available, but cached copy available here -- 1.1 GB)
• The Internet Traffic Archive:
  • SDSC HTTP, NASA HTTP, Saskatchewan HTTP – Martin Arlitt and Carey Williamson
  • France'98 World Cup HTTP – Martin Arlitt and Carey Williamson
  • Bellcore Ethernet traffic – Will Leland and Dan Wilson
  • Lawrence Berkeley Lab Internet traffic – Vern Paxson
• AOL search (no longer available from the source, but you can find mirrors on the web, e.g. at archive.org)
• Wikipedia page views
• Facebook MapReduce
• WIDE backbone access point B