Fabric和Spring以及Pool

前言

之前写了个FabricJavaPool

	implementation group: 'com.github.samyuan1990', name:'FabricJavaPool', version: '0.0.3'

最近把这个项目放到了mvn上测了一下,测试代码：
https://github.com/SamYuan1990/fabric-java-spring

测试代码逻辑

首先通过byfn 启动Fabric区块链网络
测试代码业务逻辑很简单，在8080上实现http服务器，回复"Greetings from Spring Boot! "+query a

webUI/Jemeter Java Server Blockchain Network give me "Greetings from Spring Boot! "+`query a` 3rd Round return from cache if hit. give me query a 1st Round direct link. 2nd,3rd Round Pool. 90 "Greetings from Spring Boot! "+90 webUI/Jemeter Java Server Blockchain Network

那么区别在于第一次我们不加任何cache和Pool直接连接，每次启动一个连接。
第二次我们使用Pool(FabricJavaPool)
第三次我们在最上边加个cache
笔记本电脑用jmeter，10个thread测1分钟。

测试结果

直连

惨不忍睹的直接连接。
直接连接错误了。

**46%**的错误率

Label	# Samples	Average	Median	90% Line	95% Line	99% Line	Min	Max	Error %	Throughput	Received KB/sec	Sent KB/sec
Home Page	2510	568	116	162	206	412	64	111377	46.414%	17.27770	4.17	2.00
TOTAL	2510	568	116	162	206	412	64	111377	46.414%	17.27770	4.17	2.00

带Pool

webUI/Jemeter Java Server FabricJavaPool Blockchain Network give me "Greetings from Spring Boot! "+`query a` give me query a Reused Pool connection query a 90 "Greetings from Spring Boot! "+90 webUI/Jemeter Java Server FabricJavaPool Blockchain Network

318的TPS，0 error

Label	# Samples	Average	Median	90% Line	95% Line	99% Line	Min	Max	Error %	Throughput	Received KB/sec	Sent KB/sec
Home Page	19122	29	28	42	46	51	12	63	0.000%	318.58849	59.73	38.27
TOTAL	19122	29	28	42	46	51	12	63	0.000%	318.58849	59.73	38.27

带Pool带Cache

webUI/Jemeter Java Server Cache FabricJavaPool Blockchain Network give me "Greetings from Spring Boot! "+`query a` give me query a hit 90 give me query a Reused Pool connection query a 90 "Greetings from Spring Boot! "+90 webUI/Jemeter Java Server Cache FabricJavaPool Blockchain Network

很粗暴的写了个cache

		Random r = new Random(10);
		double d2 = r.nextDouble()* 5;
		if(d2>4) {
			data = utils.QueryWithPool();
		}
		else {
			data = "90";
		}

28510.98407 TPS…

Label	# Samples	Average	Median	90% Line	95% Line	99% Line	Min	Max	Error %	Throughput	Received KB/sec	Sent KB/sec
Home Page	1710545	0	0	1	1	1	0	520	0.000%	28510.98407	5337.74	3563.87
TOTAL	1710545	0	0	1	1	1	0	520	0.000%	28510.98407	5337.74	3563.87

总结

还是要写pool和cache。
我们从46%的错误率，到了300+的tps最后到了28510TPS…
先提高了正确率，然后提升了95倍的tps.

后记

下一步我要开始给我的pool工程加cache了

Hyperledger Fabric性能测试相关文章总结（个人向）

https://blog.csdn.net/oe1019/article/details/106445904