CSE503S Performance Evaluation Study

This repository contains all of the data generated for the Performance Evaluation Study as well as the Bash and Python scripts I created to automate the testing workflow

Apache Benchmark Output

Apache Benchmark was run at 10 different concurrence levels for each of the 8 instances. Each instance has a directory, e.g. /t2Micro. Inside the folder there is a CSV file and TSV file for each concurrency level, which are a result of setting the -g and -i parameters respectively in AB. Finally, abOutput.txt is the output of running AB 10 times.

YCSB Output

Inside the directory for each instance, ycsbOutput.txt is the output of running YCSB seven times at the specified thread counts.

Scripts

Four scripts were used:

• abScript is a Bash script used to repeatedly call Apache Benchmark at increasing concurrence levels.
• abScraper.py is a Python script used to parse the response saved in abOutput.txt for each instance.
• ycsbScript is a Bash script used to repeatedly call YCSB at increasing thread count.
• ycsbScraper.py is a Python script used to parse the response saved in ycsbOutput.txt for each instance.

Other Notes

• Apache Benchmark was initially run with 5,000 requests instead of the 10,000 that were used in the study. /ab5000Tests contains the results of these tests.
• To test the consistency of Apache Benchmark, the same experiment was run three times on a T2.Large instance. The results of this is stored in /t2Large,/t2Large2 ,/t2Large3 respectively.
• ycsbResults.xlsx and abResults.xlsx store the aggregated results of all of the tests.