Taras Bobrovytsky fd1a469878 Significant improvements to benchmark report ...

- Added % change to performance regressions/improvements table
- Automatic extraction of Impala version from runtime profiles
- Execution summary row will not be printed if max time is < 100ms or < 2% of the overall runtime
- Failed queries are ignored
- First result is discarded for each query
- Geometric mean was added to summary
- Improved handling of multiple workloads in a single JSON file
- Improved handling of the case when queries are different in results and reference results
- Works well for single client runs. Additional work is needed to handle multiple client runs well.

Change-Id: Ice7b9cc4fd7502a448d35ace10fbcef183df1769
Reviewed-on: http://gerrit.sjc.cloudera.com:8080/4210
Reviewed-by: Ishaan Joshi <ishaan@cloudera.com>
Tested-by: jenkins
(cherry picked from commit c722f6b0a104df54b550978cd222a9af4d39b929)
Reviewed-on: http://gerrit.sjc.cloudera.com:8080/5250
Reviewed-by: Taras Bobrovytsky <tbobrovytsky@cloudera.com>

2014-11-13 18:54:08 -08:00

..

plugins

Update clear buffer cache plugin to only flush page cache.

2014-01-08 10:54:37 -08:00

__init__.py

Add a plugin framework for run-workload

2014-01-08 10:51:39 -08:00

create_perf_result_schema.sql

Add support for auxiliary workloads, tests, and datasets

2014-01-08 10:50:32 -08:00

perf_result_datastore.py

Added execution summary, modified benchmark to handle JSON

2014-07-25 21:06:00 -07:00

report-benchmark-results.py

Significant improvements to benchmark report

2014-11-13 18:54:08 -08:00