Thesis-ac

My bachelor thesis in Hasanuddin University as an Informatics Engineer Student.

Table of contents

• General info
• Technologies
• Run service
• Run test
• Result
  

General info

The main purpose of this thesis is to Compare the performance of microservice use gRPC and REST APIs for academic management system. So this project has 3 services:

1. KRS Service use golang.
2. Auth Service uses nodejs and also golang.
3. Payment Service use golang.

To store data in the database, I use Redis. Then use k6 to test the performance of each service and use Datadog to monitor the performance. Additionaly i also try use Jmeter to load test. All the services are deployed in different EC2 AWS as well as the test agent then use private IP to communicate.

System design

New design

Old design

Technologies

• Golang 1.17.1
• Nodejs 4.17.6
• Redis 5.0.7
• EC2 AWS
• K6
• Datadog 0.36.0
• Jmeter 5.4.1
• Linux ubuntu 20.4
  

Run service

1. KRS Service

$ IP_AUTH=localhost IP_PAYMENT=localhost IP_REDIS=localhost KRS-service/main

2. Auth Service

Use nodejs

$  cd auth-service-nodejs && npm install && IP_REDIS=localhost npm start

Use golang

$ IP_REDIS=localhost auth-service-go/main

3. Payment Service use golang.

$ IP_REDIS=localhost payment-service/main

Run test

run the cpu watch tool to watch the cpu usage.

$ cpu-watch-tool/main -p <pid server> -o <output file>

1. RUN proxy Service

$ IP_KRS=localhost IP_AUTH=localhost IP_PAYMENT=localhost  proxy/main

1. Use K6

Run DataDog agent

DOCKER_CONTENT_TRUST=1 \
sudo docker run -d \
    --name datadog \
    -v /var/run/docker.sock:/var/run/docker.sock:ro \
    -v /proc/:/host/proc/:ro \
    -v /sys/fs/cgroup/:/host/sys/fs/cgroup:ro \
    -e DD_SITE="datadoghq.com" \
    -e DD_API_KEY=<YOUR_DATADOG_API_KEY> \
    -e DD_DOGSTATSD_NON_LOCAL_TRAFFIC=1 \
    -p 8125:8125/udp \
    datadog/agent:latest

Don't forget to replace <YOUR_DATADOG_API_KEY> with your datadog api key

Run test use K6

All test files for K6 are in k6 directory. For example, if you want to test grpc-krs-create in 100 vurtual user in 30 second use the command below.

 K6_STATSD_ENABLE_TAGS=true k6 run --vus 100 --durations 30s --out statsd --tag test_run_id=1 -e IP=172.31.30.48 k6/grpc/krs/create.js

2. Use Jmeter

Run proxy service

Proxy service use to transform from rest to grpc and carry on rest to rest. So we can use jmeter to test the performance both for grpc and rest.

$ IP_KRS=127.0.0.1 IP_AUTH=127.0.0.1 IP_PAYMENT=127.0.0.1 test/main

or use the command below to automate laod test.

$ test-tools/main -j=/home/din/Downloads/apache-jmeter-5.4.1/bin/jmeter -m="test" ./jmx/1000 ./result/1000

Run test use JMeter

All jmx test file for Jmeter are in jmx directory. For example, if you want to test grpc-krs-create in 100 thread second use the command below.

jmeter.sh -n -t jmx/100/grpc-krs-create.jmx -l testresults.jtl

3. Use Locust

Still developing...

Result

Result for test use Apache JMeter

Response Time

Throughput

CPU Utilization

Result for K6 with 100 vurtual user in 30 second

Grpc

Rest

Datadog