History

Cheng Pan 68b924513c [KYUUBI #1811 ] TPC-DS benchmark expose cli args breakdown and results-dir ### _Why are the changes needed?_ Expose `breakdown` and `results-dir` as cli arg in TPC-DS benchmark tool ### _How was this patch tested?_ - [ ] Add some test cases that check the changes thoroughly including negative and positive cases if possible - [ ] Add screenshots for manual tests if appropriate - [ ] [Run test](https://kyuubi.apache.org/docs/latest/develop_tools/testing.html#running-tests) locally before make a pull request Closes #1811 from pan3793/tpcds. Closes #1811 18637ce7 [Cheng Pan] nit cc10a7ea [Cheng Pan] style 51e07398 [Cheng Pan] expose results-dir 4bc57995 [Cheng Pan] TPC-DS benchmark expose cli arg breakdown Authored-by: Cheng Pan <chengpan@apache.org> Signed-off-by: ulysses-you <ulyssesyou@apache.org>		2022-02-11 13:30:37 +08:00
..
src/main	[KYUUBI #1811 ] TPC-DS benchmark expose cli args breakdown and results-dir	2022-02-11 13:30:37 +08:00
pom.xml	[KYUUBI #1769 ][FOLLOWUP] Some cleanups for log4j properties	2022-01-28 00:58:25 +08:00
README.md	[KYUUBI #1811 ] TPC-DS benchmark expose cli args breakdown and results-dir	2022-02-11 13:30:37 +08:00

README.md

Introduction

This module includes TPC-DS data generator and benchmark tool.

How to use

package jar with following command: ./build/mvn clean package -Ptpcds -pl dev/kyuubi-tpcds -am

Data Generator

Support options:

key	default	description
db	default	the database to write data
scaleFactor	1	the scale factor of TPC-DS
format	parquet	the format of table to store data
parallel	scaleFactor * 2	the parallelism of Spark job

Example: the following command to generate 10GB data with new database tpcds_sf10.

$SPARK_HOME/bin/spark-submit \
  --class org.apache.kyuubi.tpcds.DataGenerator \
  kyuubi-tpcds_*.jar \
  --db tpcds_sf10 --scaleFactor 10 --format parquet --parallel 20

Benchmark Tool

Support options:

key	default	description
db	none(required)	the TPC-DS database
benchmark	tpcds-v2.4-benchmark	the name of application
iterations	3	the number of iterations to run
breakdown	false	whether to record breakdown results of an execution
filter	a	filter on the name of the queries to run, e.g. q1-v2.4
results-dir	/spark/sql/performance	dir to store benchmark results, e.g. hdfs://hdfs-nn:9870/pref

Example: the following command to benchmark TPC-DS sf10 with exists database tpcds_sf10.

$SPARK_HOME/bin/spark-submit \
  --class org.apache.kyuubi.tpcds.benchmark.RunBenchmark \
  kyuubi-tpcds_*.jar --db tpcds_sf10

We also support run one of the TPC-DS query:

$SPARK_HOME/bin/spark-submit \
  --class org.apache.kyuubi.tpcds.benchmark.RunBenchmark \
  kyuubi-tpcds_*.jar --db tpcds_sf10 --filter q1-v2.4

The result of TPC-DS benchmark like:

name	minTimeMs	maxTimeMs	avgTimeMs	stdDev	stdDevPercent
q1-v2.4	50.522384	868.010383	323.398267	471.6482	145.8413108576