KE-DryRun-benchmark

A benchmark tool for evaluate the performance of Ke-Query capacity

Build

mvn clean install

Start up

1. Firstly, make sure the ke_node is reachable!
2. Run the scripts

   java \
   -Dfile.dir=./samples/query_history1 \
   -Dkylin.query.node=10.1.2.134:8899 \
   -Dkylin.userName=ADMIN \
   -Dkylin.passwd=KYLIN \
   -jar target/ke-dryrun-benchmark-1.0.0.jar 

3. Benchamark data will present in the console, and store as file.

more

• Set your file.dir、node、user、passwd
• If it doesn't work, check the version of ke-dryrun-benchmark is right?

Configurations

Setting	Default Value	Required	Description
file.dir		√	CSV file path to run benchmark
concurrency	4		Number of concurrent threads
queue.size	2048		Thread pool queue size
rounds	2		Number of benchmark rounds
args.check.enabled	true		Enable argument check It will check if ke_node is reachable & check userinfo
metadata.recover.enabled	false		Enable metadata recovery
http.protocol	http		HTTP protocol
kylin.query.node		√	Kylin query node
kylin.userName	ADMIN	√	Kylin username
kylin.passwd	KYLIN	√	Kylin password
report.output.dir	.		Report output directory
logging.level.io.kyligence	info		logging level

enjoy the benchmark travel

Tools

SqlExtractTool

help extract sql from txt to executable csv file that benchmark works

usages

# unzip jar
unzip ke-dryrun-benchmark-1.0.0.jar

# execute java file by setting classpath & params
java -cp BOOT-INF/classes/:"BOOT-INF/lib/*" io.kyligence.benchmark.tool.SqlExtractTool [sql.txt.filepath] [csv.outputPath] [ke_node] [project] [username] [password]

params desc

param	Required	description
sql.txt.filepath	√	csv or txt filled with sql
csv.outputPath	√	output csv file path: e.g. benchmark.csv
ke_node	√	ke node e.g. ip:port
project	√	projectName that can execute sql
username	√	username
passwd	√	password

Next to Do

[ ] more metric info such as 90line,95line, median [ ] integrate with feishu documents [ ] fix some http errors