Changes between Version 9 and Version 10 of PlasticSlices/BaselineEvaluation/Baseline3Details


Ignore:
Timestamp:
06/07/11 11:27:02 (12 years ago)
Author:
Josh Smift
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • PlasticSlices/BaselineEvaluation/Baseline3Details

    v9 v10  
    66
    77The Stanford FlowVisor was suffering from significant instabilities during this baseline. We were originally planning to use of-planet3.stanford.edu and of-planet4.stanford.edu in three experiments, but those plnodes were VMs, and were frequently unreachable. We switched to using of-planet1 and ofplanet2, which are on dedicated hardware, for round 1, but the FlowVisor problems also left them unreachable too much of the time, so we dropped the Stanford plnodes from this baseline entirely for rounds 2 - 6. The tables of client/server pairs are for rounds 2 - 6.
     8
     9We ran a 'screen' process for each slice, and connected to each plnode in a window in that 'screen' process; we then logged each session using the 'screenlog' feature, which saves files in the same format as the 'typescript' command, i.e. including all keystrokes entered, terminal escape sequences, etc. The raw logs are at http://www.gpolab.bbn.com/plastic-slices/baseline-logs/baseline-3/, with a subdirectory for the logs for each round.
    810
    911= plastic-101 =