大数据基础之Benchmark（1）HiBench

HiBench 7
官方：https://github.com/intel-hadoop/HiBench

一 简介

HiBench is a big data benchmark suite that helps evaluate different big data frameworks in terms of speed, throughput and system resource utilizations. It contains a set of Hadoop, Spark and streaming workloads, including Sort, WordCount, TeraSort, Sleep, SQL, PageRank, Nutch indexing, Bayes, Kmeans, NWeight and enhanced DFSIO, etc. It also contains several streaming workloads for Spark Streaming, Flink, Storm and Gearpump.

There are totally 19 workloads in HiBench.

Supported Hadoop/Spark/Flink/Storm/Gearpump releases:

Hadoop: Apache Hadoop 2.x, CDH5, HDP
Spark: Spark 1.6.x, Spark 2.0.x, Spark 2.1.x, Spark 2.2.x
Flink: 1.0.3
Storm: 1.0.1
Gearpump: 0.8.1
Kafka: 0.8.2.2

二 spark sql测试

1 download

$ wget https://github.com/intel-hadoop/HiBench/archive/HiBench-7.0.tar.gz
$ tar xvf HiBench-7.0.tar.gz
$ cd HiBench-HiBench-7.0

2 build

1）build all

$ mvn -Dspark=2.1 -Dscala=2.11 clean package

2）build hadoopbench and sparkbench

$ mvn -Phadoopbench -Psparkbench -Dspark=2.1 -Dscala=2.11 clean package

3）only build spark sql

$ mvn -Psparkbench -Dmodules -Psql -Dspark=2.1 -Dscala=2.11 clean package

3 prepare

$ cp conf/hadoop.conf.template conf/hadoop.conf
$ vi conf/hadoop.conf

$ cp conf/spark.conf.template conf/spark.conf
$ vi conf/spark.conf

$ vi conf/hibench.conf
# Data scale profile. Available value is tiny, small, large, huge, gigantic and bigdata.
# The definition of these profiles can be found in the workload's conf file i.e. conf/workloads/micro/wordcount.conf
hibench.scale.profile bigdata

4 run

sql测试分为3种：scan/aggregation/join

$ bin/workloads/sql/scan/prepare/prepare.sh
$ bin/workloads/sql/scan/spark/run.sh

具体配置位于conf/workloads/sql/scan.conf
prepare之后会在hdfs的/HiBench/Scan/Input下生成测试数据，在report/scan/prepare/下生成报告
run之后会在report/scan/spark/下生成报告，比如monitor.html，在hive的default库下可以看到测试数据表

$ bin/workloads/sql/join/prepare/prepare.sh
$ bin/workloads/sql/join/spark/run.sh

$ bin/workloads/sql/aggregation/prepare/prepare.sh
$ bin/workloads/sql/aggregation/spark/run.sh

依此类推

 

如果prepare时报错内存溢出

尝试修改

$ vi bin/functions/workload_functions.sh
local CMD="${HADOOP_EXECUTABLE} --config ${HADOOP_CONF_DIR} jar $job_jar $job_name $tail_arguments"

格式：hadoop jar <jarName> <youClassName> -D mapreduce.reduce.memory.mb=5120 -D mapreduce.reduce.java.opts=-Xmx4608m <otherArgs>

发现不能生效，尝试增加map数量

$ vi bin/functions/hibench_prop_env_mapping.py:
NUM_MAPS="hibench.default.map.parallelism",

$ vi conf/hibench.conf
hibench.default.map.parallelism 5000

 

参考：
https://github.com/intel-hadoop/HiBench/blob/master/docs/build-hibench.md
https://github.com/intel-hadoop/HiBench/blob/master/docs/run-sparkbench.md