测试环境:
使用机器4台 ,环境配置如下
项目 | 数值 |
---|---|
网卡 | 10000Mb/s |
磁盘 | 10块 * 2T * 4节点 HDD |
副本数 | 3 |
内存 | 32G12块 4节点 |
HDFS集群组成 | Name Node2 + Data Node4 |
HDFS集群版本 | 2.7.5 |
CPU | Intel® Xeon® Gold 6230R 52核 * 4节点 |
测试工具包 | hadoop-mapreduce-client-jobclient-2.7.2-transwarp-7.0.1-tests.jar |
HDFS具体服务配置参数:
1、NameNode:GC_OPTS -Xmx 24000m
2、DataNode:GC_OPTS Xmx 8192m
3、dfs.namenode.handler.count 100
4、dfs.datanode.handler.count 100
测试场景:
(1)向HDFS文件系统中写入数据,500个文件,每个文件 4195 MB。总数据2T。
指令:
time yarn jar hadoop-mapreduce-client-jobclient-2.7.2-transwarp-7.0.1-tests.jar TestDFSIO -write -nrFiles 500 -fileSize 4195 >>./TestDFSIO_500_read.log 2>&1
第一次结果输出:
写文件
文件数量:500个
处理的总字节数:2097500
吞吐量mb/s:14.09
平均IO速率mb/sec: 14.99
IO rate std deviation:4.22
测试执行时间秒: 1261.49
第二次结果输出:
----- TestDFSIO ----- : write
Date & time: Tue Feb 08 16:37:07 CST 2022
Number of files: 500
Total MBytes processed: 2097500
Throughput mb/sec: 10.92
Average IO rate mb/sec: 13.1
IO rate std deviation: 7.46
Test exec time sec: 2260.96
第三次结果输出:
----- TestDFSIO ----- : write
Date & time: Sat Feb 12 10:32:36 CST 2022
Number of files: 500
Total MBytes processed: 2097500
Throughput mb/sec: 11.92
Average IO rate mb/sec: 13.45
IO rate std deviation: 5.43
Test exec time sec: 1539.11
(2)读取数据测试
指令:
time yarn jar hadoop-mapreduce-client-jobclient-2.7.2-transwarp-7.0.1-tests.jar TestDFSIO -read -nrFiles 500 -fileSize 4195 >>./TestDFSIO_500_read.log 2>&1
第一次结果输出:
----- TestDFSIO ----- : read
Date & time: Tue Feb 08 16:58:32 CST 2022
Number of files: 500
Total MBytes processed: 2097500
Throughput mb/sec: 19.92
Average IO rate mb/sec: 29.83
IO rate std deviation: 20.08
Test exec time sec: 893.61
第二次结果输出:
----- TestDFSIO ----- : read
Date & time: Tue Feb 08 17:13:35 CST 2022
Number of files: 500
Total MBytes processed: 2097500
Throughput mb/sec: 28.54
Average IO rate mb/sec: 31.91
IO rate std deviation: 11.02
Test exec time sec: 637.79
第3次结果输出:
----- TestDFSIO ----- : read
Date & time: Sat Feb 12 10:56:41 CST 2022
Number of files: 500
Total MBytes processed: 2097500
Throughput mb/sec: 27.12
Average IO rate mb/sec: 30.03
IO rate std deviation: 10.12
Test exec time sec: 666.06
(3)测试写文件性能测试规格 263157个1MB文件,数据量256G
指令:
time yarn jar hadoop-mapreduce-client-jobclient-2.7.2-transwarp-7.0.1-tests.jar TestDFSIO -write -nrFiles 263157 -fileSize 1MB >> ./TestDFSIO_500_read.log 2>&1
第一次结果输出:
----- TestDFSIO ----- : write
Date & time: Wed Feb 09 01:20:08 CST 2022
Number of files: 263157
Total MBytes processed: 263157
Throughput mb/sec: 3.3
Average IO rate mb/sec: 4.3
IO rate std deviation: 1.34
Test exec time sec: 13930.2
第二次结果输出:
----- TestDFSIO ----- : write
Date & time: Thu Feb 10 19:12:51 CST 2022
Number of files: 263157
Total MBytes processed: 263157
Throughput mb/sec: 1.29
Average IO rate mb/sec: 4
IO rate std deviation: 1.63
Test exec time sec: 18124.23
第3次结果输出:
----- TestDFSIO ----- : write
Date & time: Sat Feb 12 18:45:33 CST 2022
Number of files: 263157
Total MBytes processed: 263157
Throughput mb/sec: 3.61
Average IO rate mb/sec: 4.36
IO rate std deviation: 1.29
Test exec time sec: 13341.98
(4)测试读文件性能测试
指令:
time yarn jar hadoop-mapreduce-client-jobclient-2.7.2-transwarp-7.0.1-tests.jar TestDFSIO -read -nrFiles 263157 -fileSize 1MB >> ./TestDFSIO_500_read.log 2>&1
第一次结果输出:
----- TestDFSIO ----- : read
Date & time: Wed Feb 09 17:40:26 CST 2022
Number of files: 263157
Total MBytes processed: 263157
Throughput mb/sec: 11.55
Average IO rate mb/sec: 39.29
IO rate std deviation: 43.22
Test exec time sec: 13377.81
第二次结果输出:
----- TestDFSIO ----- : read
Date & time: Thu Feb 10 04:04:08 CST 2022
Number of files: 263157
Total MBytes processed: 263157
Throughput mb/sec: 11.33
Average IO rate mb/sec: 39.21
IO rate std deviation: 43.54
Test exec time sec: 13979.92
第三次结果输出:
----- TestDFSIO ----- : read
Date & time: Sun Feb 13 02:32:27 CST 2022
Number of files: 263157
Total MBytes processed: 263157
Throughput mb/sec: 12.66
Average IO rate mb/sec: 38.7
IO rate std deviation: 45.33
Test exec time sec: 12911.45
测试数据归纳
系统级指标名称 | 指标名称 | 数据量 | 测试内容 | 测试结果 |
---|---|---|---|---|
吞吐量规格 | 写吞吐量(MB/Node/Sec) | 2T(500个4195 MB文件) | hadoop-mapreduce-client-jobclient-2.7.2-transwarp-7.0.1-tests.jar | 第一次:3.52 MB/Node/Sec ,第二次:2.73 MB/Node/Sec ,第三次: 2.98 MB/Node/Sec |
吞吐量规格 | 读吞吐量(MB/Node/Sec) | 2T(500个4195 MB文件 | hadoop-mapreduce-client-jobclient-2.7.2-transwarp-7.0.1-tests.jar | 第一次:4.98 MB/Node/Sec ,第二次:7.10 MB/Node/Sec ,第三次:6.78 MB/Node/Sec |
吞吐量规格 | 写文件(次/秒/NN) | 256G(263157个1MB文件) | hadoop-mapreduce-client-jobclient-2.7.2-transwarp-7.0.1-tests.jar | 第一次:13377.81 次/秒/NN,第二次:13979.92 次/秒/NN,第三次:12911.45 次/秒/NN |
吞吐量规格 | 读文件(次/秒/NN) | 256G(263157个1MB文件) | hadoop-mapreduce-client-jobclient-2.7.2-transwarp-7.0.1-tests.jar | 第一次:13930.2 次/秒/NN,第二次:18124.23 次/秒/NN,第三次:13341.98 次/秒/NN |