Three routes are tested :
- /big : return a big json object (354Ko)
- /small : return a small json object (2Ko)
- /hello : return
{hello : 'world'}
Data sent back by both /big and /small routes are real-world data (anonymized) corresponding some of our real application responses.
Updated 19/09/2017
Node : 8.5.0
Pc : Intel(R) Core i7-6500U CPU 2.50GHZ 2.59GHZ 8.00Go RAM
OS : Windows 10
3 runs : only the results of third run are kept.
express 4.15.4
npm run test-perf-express
> frameworks-benchmark@0.0.0 test-perf-express C:\node-projects\frameworks-benchmark
> autocannon -c 100 -d 5 localhost:3000/big && autocannon -c 100 -d 5 localhost:3000/small && autocannon -c 100 -d 5 localhost:3000/
hello
Running 5s test @ http://localhost:3000/big
100 connections
Stat Avg Stdev Max
Latency (ms) 169.47 50.65 379
Req/Sec 577.6 23.07 601
Bytes/Sec 98.6 MB 4.59 MB 105 MB
3k requests in 5s, 493 MB read
Running 5s test @ http://localhost:3000/small
100 connections
Stat Avg Stdev Max
Latency (ms) 9.15 3.32 49
Req/Sec 10336.4 1511.37 11119
Bytes/Sec 8.89 MB 1.38 MB 9.96 MB
52k requests in 5s, 44.1 MB read
Running 5s test @ http://localhost:3000/hello
100 connections
Stat Avg Stdev Max
Latency (ms) 8.12 2.95 43
Req/Sec 11610.4 832.47 12383
Bytes/Sec 2.16 MB 156 kB 2.36 MB
58k requests in 5s, 10.9 MB read
fastify 0.27
$ npm run test-perf-fastify
> frameworks-benchmark@0.0.0 test-perf-fastify C:\node-projects\frameworks-benchmark
> autocannon -c 100 -d 5 localhost:3000/big && autocannon -c 100 -d 5 localhost:3000/bigschema && autocannon -c 100 -d 5 localhost:3
000/small && autocannon -c 100 -d 5 localhost:3000/hello
Running 5s test @ http://localhost:3000/big
100 connections
Stat Avg Stdev Max
Latency (ms) 194.49 13.32 238
Req/Sec 507.8 26.2 533
Bytes/Sec 86 MB 4.59 MB 92.3 MB
3k requests in 5s, 434 MB read
Running 5s test @ http://localhost:3000/bigschema
100 connections
Stat Avg Stdev Max
Latency (ms) 171.4 10.6 215
Req/Sec 580.21 39.6 600
Bytes/Sec 70.7 MB 5.45 MB 75.5 MB
3k requests in 5s, 345 MB read
Running 5s test @ http://localhost:3000/small
100 connections
Stat Avg Stdev Max
Latency (ms) 4.04 1.55 37
Req/Sec 22388.8 1850.55 23535
Bytes/Sec 18.6 MB 1.57 MB 19.9 MB
112k requests in 5s, 91.2 MB read
Running 5s test @ http://localhost:3000/hello
100 connections
Stat Avg Stdev Max
Latency (ms) 2.96 1.45 39
Req/Sec 28804.8 2243.12 30671
Bytes/Sec 4.29 MB 360 kB 4.72 MB
144k requests in 5s, 21.5 MB read
- Fastify is much faster then Express in 'small' and 'hello' tests (roughly x2 performance)
- However concerning the 'big' test, Express is sligthly better.
- With big response, Fastify can only match Express performance by using schema validation (route /bigschema in the tests) which allows to enable faster Json stringification