Code

Code resides in assign+ branch of the repository, in assign+ folder. There are three files:

• assign+
  • PURPOSE: reads resource requirements and available resources and finds an allocation with minimum interswitch bw use
  • INPUT: two command line arguments ptopfile and topfile. These files are usually produced in the process of resource assignment by assign_wrapper, now called mapper. There are a few other command line options, that should be self-explanatory when you run the code. If you want very verbose output run it with -d.
  • OUTPUT: textual output denoting nodes and links and their assignments that can be passed further to resource allocation
• getptopdetail.pl
  • PURPOSE: parse a ptopfile into something human-readable. Assign+ also uses this output.
  • INPUT: ptopfile
  • OUTPUT: human-readable summary of ptopfile
• gettopdetail.pl
  • PURPOSE: parse a topfile into something human-readable. Assign+ also uses this output.
  • INPUT: topfile
  • OUTPUT: human-readable summary of topfile

Testing

Testing code resides in assign+ branch of the repository, in assign+/test folder. There are several files that jointly perform tests that compare performance of assign and assign+.

• removefixed.pl
  • PURPOSE: Remove fixed nodes from a topfile
  • INPUT: topfile
  • OUTPUT: topfile in correct format minus lines that talk about fixed nodes
• makefixed.pl
  • PURPOSE: Make a topfile with fixed nodes taken from the output of assign or assign+
  • INPUT: output-file-of-assign-or-assign+ topfile
  • OUTPUT: topfile in correct format plus lines that enforce fixed nodes and fixed interfaces for links
• parseinfo.pl
  • PURPOSE: measure performance of assign or assign+
  • INPUT: a file containing output of assign or assign+
  • OUTPUT: success or failure of allocation, time the allocation took, number of nodes and interswitch links allocated
• runtests.pl
  • PURPOSE: run a number of tests with assign and assign+. The tests we run are the following:
    • t1. assign with given ptopfile and topfile
    • t2. assign+ with given ptopfile and topfile
    • t3. assign with given ptopfile and modified topfile to remove fixed nodes
    • t4. assign+ with given ptopfile and modified topfile to remove fixed nodes
    • t5. assign with given ptopfile and modified topfile to remove fixed nodes, and then include fixed nodes and fixed interfaces from the output of assign+ in step t4
  • INPUT: a file containing lines with ptopfile and topfile
  • OUTPUT: file testresults that contains performance of these two allocation algorithms for each line of input
• processresults.pl
  • PURPOSE: process the results file generated in the previous step
  • INPUT: name of the results file
  • OUTPUT: statistics about successes and failures of assign and assign+. The code also produces files that contain inputs that lead to following potentially anomalous cases:
    • checktests - assign+ failed and assign didn't, i.e. t1 was a success and t2 was a failure
    • failedtests - both assign+ and assign failed, i.e. both t1 and t2 were failures
    • weirdtests - assign+ succeeded and assign didn't, i.e. t1 was a failure and t2 and t5 were a success
    • impossibletests - assign+ succeeded and assign didn't with fixed nodes from output of assign+, i.e. t1 was a failure, t2 was a success and t5 was a failure. This most often happens because assign cannot deal properly with fixed interfaces on links.

Additionally, the folder contains a version of assign I have used in testing, that will run on boss. It also contains two sets of test cases:

• expinfotests - all attempted allocations from start of DeterLab until Spring 2013
• expfailedtests - all failed allocations roughly from Jan 2012 until Spring 2013

Performance

Expinfotests

There were total of 109,326 tests. Historically, in real operation some of these allocations succeeded and some failed.