This benchmark aims to provide a naive gauge of performance on a typical simple production REST API (running with net/http and Express.js). It is NOT meant to demonstrate the maximum performance achievable from Go and Node.js.
Receive a pair of numbers and return their sum as the result.
Request is a JSON message containing two numbers:
{
"a": 123,
"b": 456
}Response is a JSON message containing the result:
{
"result": 579
}Testing Rig Specs
OS: Ubuntu 20.04 LTS x86_64
Kernel: 5.4.0-37-generic
CPU: Intel i5-8350U (8 CPUs) @ 3.600GHz
Memory: 15871MiB
Benchmark Tool Apache's ab tool was used with the following options:
$ ab -p data.json -T application/json -c 1000 -n 50000 http://localhost:<PORT>/Go
- net/http
- chi-go for routing
Node.js
- Express.js web application framework
- Disabled Etag and X-Powered-By in HTTP response headers
- PM2 for scaling with cluster mode
Go
Concurrency Level: 1000
Time taken for tests: 2.893 seconds
Complete requests: 50000
Failed requests: 0
Total transferred: 6700000 bytes
Total body sent: 8200000
HTML transferred: 850000 bytes
Requests per second: 17280.86 [#/sec] (mean)
Time per request: 57.867 [ms] (mean)
Time per request: 0.058 [ms] (mean, across all concurrent requests)
Transfer rate: 2261.36 [Kbytes/sec] received
2767.64 kb/s sent
5029.00 kb/s total
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 28 6.3 28 54
Processing: 10 29 6.3 29 56
Waiting: 0 18 7.4 16 45
Total: 38 57 2.9 57 84
Percentage of the requests served within a certain time (ms)
50% 57
66% 57
75% 58
80% 59
90% 60
95% 62
98% 64
99% 66
100% 84 (longest request)
Node.js
Without PM2 (single thread):
Concurrency Level: 1000
Time taken for tests: 6.766 seconds
Complete requests: 50000
Failed requests: 0
Total transferred: 7900000 bytes
Total body sent: 8200000
HTML transferred: 800000 bytes
Requests per second: 7390.41 [#/sec] (mean)
Time per request: 135.311 [ms] (mean)
Time per request: 0.135 [ms] (mean, across all concurrent requests)
Transfer rate: 1140.32 [Kbytes/sec] received
1183.62 kb/s sent
2323.94 kb/s total
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 15 93.2 5 1023
Processing: 46 119 18.9 116 368
Waiting: 18 89 17.7 88 346
Total: 46 134 97.3 121 1391
Percentage of the requests served within a certain time (ms)
50% 121
66% 122
75% 124
80% 125
90% 138
95% 167
98% 252
99% 260
100% 1391 (longest request)
With PM2 (multiple threads):
Concurrency Level: 1000
Time taken for tests: 6.931 seconds
Complete requests: 50000
Failed requests: 0
Total transferred: 7900000 bytes
Total body sent: 8200000
HTML transferred: 800000 bytes
Requests per second: 7213.76 [#/sec] (mean)
Time per request: 138.624 [ms] (mean)
Time per request: 0.139 [ms] (mean, across all concurrent requests)
Transfer rate: 1113.06 [Kbytes/sec] received
1155.33 kb/s sent
2268.39 kb/s total
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 3 5.3 1 37
Processing: 17 134 20.2 130 216
Waiting: 15 132 19.8 129 216
Total: 53 137 18.8 132 217
Percentage of the requests served within a certain time (ms)
50% 132
66% 142
75% 147
80% 152
90% 163
95% 174
98% 188
99% 194
100% 217 (longest request)
Speed of handling requests
For this Sum API, Go appears to handle the request around 2x faster than Node.js.
CPU load
Go: all CPUs running at around 20%-40% load
Node.js (single thread): one CPU running at around 90% load
Node.js (multiple threads w/PM2): all CPUs running at around 80%-90% load
Others
Interestingly, Node.js with a single thread handled requests at about the same speed as Node.js with PM2. Perhaps Node.js with PM2 will show a greater advantage for longer-running and more CPU intensive tasks.