Awesome
PlainTextResultsParser for FrameWorkBenchmark
Summary
Provide a plaintext results FrameWorkBenchmark json parser based on python prettytable, Pros:
- No need to copy files from server
- Easy to visualize data instantly and automatically
- More flexible, we could select interested data and even compare between runs
Demo：
python FrameWorkBenchmark_parser.py --data latencyAvg totalRequests --files netty-async/results.json.1 netty-async/results.json.2
+---------------------------------------------------------------+
| Type: plaintext, Result: latencyAvg |
+---------------------------+---------------------+-------------+
| pipelineConcurrencyLevels | netty-async-virtual | netty-async |
+---------------------------+---------------------+-------------+
| 4 | 71.33us | 96.84us |
| 8 | 113.10us | 214.25us |
| 16 | 299.17us | 498.47us |
| 32 | 530.01us | 0.90ms |
| 256 | 4.59ms | 8.48ms |
| 1024 | 19.88ms | 34.68ms |
+---------------------------+---------------------+-------------+
+-------------------------------------------------------+
| Type: json, Result: latencyAvg |
+-------------------+---------------------+-------------+
| concurrencyLevels | netty-async-virtual | netty-async |
+-------------------+---------------------+-------------+
| 4 | 68.86us | 73.23us |
| 8 | 85.92us | 96.26us |
| 16 | 104.07us | 119.98us |
| 32 | 135.47us | 209.89us |
| 64 | 257.47us | 416.03us |
| 128 | 516.78us | 845.06us |
| 256 | 1.04ms | 1.69ms |
| 512 | 2.07ms | 3.39ms |
+-------------------+---------------------+-------------+
Usage
Setup via pip3 install prettytable, list results json files with --files and interested data with --data.
Note:
- The result files must be formatted by FrameWorkBenchmark
- Make sure different result files are from the same stress level when comparing them