open-estuary/appbenchmark

ARM64-based Application Performance Benchmark Tests

• Introduction
• Applibcation Benchmark Test
• Ansible roles
• Others

Estuary Application Benchmark Test

Introduction

It is to maintain application benchmark test scripts including:

• Scripts which setup benchmark test server and client
• Scripts which perform benchmark test

Bebfore running Ansible, it is necessary to create new account on target servers. Then the later test could use this account to login in target servers without password. As for how to create new accounts automatically, please refer to createuser.sh

It is required to write new benchmark scripts based on ansible framework. In addition, some existing old scripts might be changed to ansible framework in the future. As for test cases based on ansible framework, it might be necessary to change some configurations before setting up benchmark test based on real test environment, such as: ansible/hosts which specifies test hosts, and ansible/group_vars/* which define variables for this test.

Applibcation Benchmark Test

Usually it is only necessary to execute following commands in order to perform benchmark test for each application:

• ./setup.sh to setup client and server side;
• ./run_{test/xxx}.sh to perform benchmark test on client side

Currently the following applications have been supported so far :

• Basic Test

  • CPU
    • SpecInt20016
  • IO
    • FIO
    • Sysbench
  • Memory
    • LmBench    - Stream
  • Network
    • Iperf/Qperf/NetPerf
  • JVM
    • SpecJVM2008

• MySQL(Percona Server)

  • Test Case 1(1 large instance test which focus on latency)

• AliSQL

  • Test Case 1(200 small instances test)
  • Test Case 2(5 middle instances test)

• Cassandra

• Ceph

• MariaDB

• PostgreSQL

• MongoDB

• Redis

• Solr

• Hadoop

• Flume

• ARM64 Based E-Commerce Application Solutions

  • E-Commerce-SpringCloud-Microservice Solution
  • E-Commerce-Kubernetes Solution

• Docker Networks

  • Calico
  • Flannel  - Weave

• BigData

Ansible Roles

Usually the Ansible role need to be configured and could be reused for difference application benchmark tests (or ansible playbook).

Currently the following applications ansible roles have been supported:

Application Name	Notes
spec_cpu2006	CPU Test Role
FIO	FIO Test Role
Sysbench	Sysbench Role
LmBench	LmBench
Stream	Stream Test Role
Iperf/Qperf/NetPerf	Network Performance Test Role
MySQL	MySQL Test Role
Percona Server	Percona Server Test Role
Redis	Redis Test Role
twemproxy	Twemproxy Test role
estuaryrepo	Provision Estuary Repository
nginx	Nginx Web Server Role
nginx-reverse-proxy	Nginx Load Balancer Role
wrk	WRK Test Role
php-fpm	PHP-FPM Role
solr	Single Role Role
solrcloud	Single Cloud Role
jmeter	Apache JMeter Role
docker	Install Docker Role
docker_net	Docker Network(including Flannel, Weave) Roles
irqbalance	Set Irqbalance Role
openjdk	Install OpenJDK Role
weighttp	Weighthttp Role
kafka	Apache Kafka Role
flume	Flume Role
cart	E-commerce Cart MicroService Role
search	E-commerce Search MicroService Role
order	E-commerce Order MicroService Role
zuul-apigateway	Sring Cloud Zuul Role
eureka-service-discovery	Spring Cloud Eureka Role
createuser	Create new user account role
mount	Mount new devices role
syscfg	Provision Kernel Parameters role
bigdata	Bigdata Install Test role

Others

• As for how to write new test case for applications, please refer to how_to_write_newcase