workload-analyzer

Workload Analyzer

This module analyzes a workload after it is run.

(For now, it only supports OpenWhisk activations.)

How to use Workload Analyzer

1. The Synthetic Workload Invoker should be run first.
2. Then run the Workload Analyzer:

python WorkloadAnalyzer.py -r -p

3. Certain features can be controlled by input arguments:
   1. -v or --verbose: prints the detailed test data
   2. -p or --plot: plots the test results
   3. -s or --save_plot: save test result plots (this option should be used with -p)
   4. -a or --archive: archive the test results in an pickle file (in the data_archive directory)
   5. -c or --capacity_factor: returns the capacity factor for functions in the workload (stored as JSON in workload-analyzer/capacity_factors.json)
   6. -o or --override_testname: this option is followed by the new test name. Allows assigning new names to tests, which is specifically useful for archivning with desired names.
   7. -r or --read_results: also gather the results of function invocations
4. Analysis logs can be found in ../logs/WA.log.

Required Packages (beyond standard libraries)

• matplotlib
• pandas
• seaborn

Tested Environments

Environment/Tool	Tested Version
Python	3.6.5 & 3.6.8
OS	Ubuntu 16.04.4 LTS

Important Note:

The wskutil.py script has been taken from the OpenWhisk project (https://github.com/apache/openwhisk/blob/master/tools/admin/wskutil.py) with no modifications. It holds the original Apache 2.0 license.