Kafka 自带测试脚本进行性能测试

压力测试

用Kafka官方自带的脚本，对Kafka进行压测。Kafka压测时，可以查看到哪个地方出现了瓶颈（CPU，内存，网络IO）。一般都是网络IO达到瓶颈。

producer 压力测试

bin/kafka-producer-perf-test.sh  --topic test --record-size 100 --num-records 100000 --throughput 1000 --producer-props bootstrap.servers=dw-node01:9092,dw-node02:9092,dw-node03:9092

• record-size 设置单条信息大小 单位：字节
• num-records 设置总共多少条信息
• throughput 设置每秒产生多少条信息

测试结果

5001 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 3.0 max latency.
5002 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 2.0 max latency.
5001 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 6.0 max latency.
5002 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 4.0 max latency.
5002 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 13.0 max latency.
5001 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 3.0 max latency.
5002 records sent, 1000.0 records/sec (0.10 MB/sec), 0.9 ms avg latency, 2.0 max latency.
5003 records sent, 1000.4 records/sec (0.10 MB/sec), 0.9 ms avg latency, 2.0 max latency.
5002 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 2.0 max latency.
5002 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 2.0 max latency.
5001 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 2.0 max latency.
5002 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 14.0 max latency.
5001 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 8.0 max latency.
5002 records sent, 1000.2 records/sec (0.10 MB/sec), 0.9 ms avg latency, 2.0 max latency.
100000 records sent, 999.890012 records/sec (0.10 MB/sec), 1.21 ms avg latency, 539.00 ms max latency, 1 ms 50th, 1 ms 95th, 2 ms 99th, 69 ms 99.9th.

本例中一共写入10w条消息，每秒向Kafka写入了0.10MB的数据，平均是999.89条消息/秒，每次写入的平均延迟为0.10毫秒，最大的延迟为539毫秒…

Consumer 压力测试

Consumer的测试，如果这四个指标（IO，CPU，内存，网络）都不能改变，考虑增加分区数来提升性能

bin/kafka-consumer-perf-test.sh --zookeeper dw-node01:2181 --topic test --fetch-size 10000 --messages 10000000 --threads 1

• zookeeper 指定zookeeper的链接信息

• topic 指定topic的名称

• fetch-size 指定每次fetch的数据的大小

• messages 总共要消费的消息个数

测试结果

start.time, end.time, data.consumed.in.MB, MB.sec, data.consumed.in.nMsg, nMsg.sec
2020-07-20 12:01:03:058, 2020-07-20 12:01:03:902, 9.5367, 11.2995, 100000, 118483.4123

开始测试时间，测试结束数据，最大吞吐率9.5367MB/s，平均每秒消费11.2995MB/s，最大每秒消费100000条，平均每秒消费118483.4123条。