node-vs-golang

In this project what is tried is to compare the effectiveness of a Node and Golagitng server with an overload test.

Folder structure

docker-compose.yml                 # Containers
    ├─── docs                      # Documentation
    ├─── api-golang                # API Rest with Golang
    ├─── api-node                  # API Rest with NodeJS
    └─── bench.sh                  # Script for launch test

Requirements

• Docker y Docker-Compose
• Bash 5.0
• gnuplot

Tests

The tests configured in the code are:

Test	Requests	Concurrents
0	100	1
1	1000	100
2	10000	1000

To modify or add more tests, you must go to the file bench.sh and in the first lines they show the structure of the tests:

    # TEST 0
    declare -A test0=(
        [requests]=100
        [concurrent]=1
        [name]='test0'
    )

The field name is important because save de data under the name.

Init

1. Clone the repository:

    git clone https://github.com/JaimeRC/node-vs-golang.git
   

2. Go to project folder:

    cd node-vs-golang
   

3. Init docker-compose:

    docker-compose up
    
    Starting apigolang ... done
    Starting apinode   ... done
    Attaching to apinode, apigolang
    apigolang     | 2020/10/21 13:44:14 Running on port 8080
    apinode       | 
    apinode       | > api-node@1.0.0 start /src
    apinode       | > node ./server/index.js
    apinode       | 
    apinode       | Server listening to  7070
   

   If you add the parameter -d it initializes in the background.

4. In another terminal, start the tests:

   • Node Server:

       bash bench.sh node
       
       Benching: http://localhost:7070/api
     

   • Golang Server:

       bash bench.sh golang
       
       Benching: http://localhost:8080/api
     

5. After the tests, three files are created per server:

   • Description the tests in output_node.txt (only the first is shown):

       Benching: http://localhost:7070/api
       '1' concurrent user doing '100' page hits
       This shows you how well the web-server will handle a simple load of '1' user doing a number of page loads.
       
       This is ApacheBench, Version 2.3 <$Revision: 1843412 $>
       Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
       Licensed to The Apache Software Foundation, http://www.apache.org/
       
       Benchmarking localhost (be patient).....done
       
       Server Software:        
       Server Hostname:        localhost
       Server Port:            7070
       
       Document Path:          /api/test
       Document Length:        Variable
       
       Concurrency Level:      1
       Time taken for tests:   0.143 seconds
       Complete requests:      100
       Failed requests:        0
       Non-2xx responses:      100
       Keep-Alive requests:    100
       Total transferred:      79900 bytes
       HTML transferred:       900 bytes
       Requests per second:    697.27 [#/sec] (mean)
       Time per request:       1.434 [ms] (mean)
       Time per request:       1.434 [ms] (mean, across all concurrent requests)
       Transfer rate:          544.06 [Kbytes/sec] received
       
       Connection Times (ms)
                     min  mean[+/-sd] median   max
       Connect:        0    0   0.0      0       0
       Processing:     1    1   0.6      1       6
       Waiting:        1    1   0.6      1       6
       Total:          1    1   0.6      1       6
       
       Percentage of the requests served within a certain time (ms)
         50%      1
         66%      1
         75%      1
         80%      1
         90%      2
         95%      2
         98%      4
         99%      6
        100%      6 (longest request)
       
       =============================================================================================================
     

   • Graph with the results per test in node_test0.png (only the first is shown):