The repository is a high-performance tech stack based on Spring Boot that using Netty instead of Spring Web MVC / Spring Webflux.
KVM, Ubuntu 16.04, 2048 MB RAM, 2 CPU
Run: $ java -Xmx1024M -server -jar application.jar
Test: $ wrk -t12 -c400 -d30s --latency http://host/api/user/current
Running 30s test @ http://host/api/user/current
12 threads and 400 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 54.66ms 64.39ms 996.17ms 92.02%
Req/Sec 744.51 140.25 3.08k 85.77%
Latency Distribution
50% 42.26ms
75% 65.72ms
90% 105.89ms
99% 275.51ms
265240 requests in 30.06s, 53.93MB read
Requests/sec: 8822.80
Transfer/sec: 1.79MB
Running 30s test @ http://host/api/user/current
12 threads and 400 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 66.80ms 121.30ms 1.94s 91.37%
Req/Sec 0.87k 327.28 6.07k 91.60%
Latency Distribution
50% 27.18ms
75% 77.43ms
90% 172.98ms
99% 505.67ms
303177 requests in 30.04s, 51.47MB read
Socket errors: connect 0, read 0, write 0, timeout 148
Requests/sec: 10090.98
Transfer/sec: 1.71MB
Running 30s test @ http://host/api/user/current
12 threads and 400 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 64.22ms 101.43ms 2.00s 89.74%
Req/Sec 0.88k 332.80 7.93k 93.13%
Latency Distribution
50% 27.41ms
75% 77.47ms
90% 168.43ms
99% 432.62ms
306800 requests in 30.10s, 52.08MB read
Socket errors: connect 0, read 0, write 0, timeout 36
Requests/sec: 10193.27
Transfer/sec: 1.73MB
Running 30s test @ http://host/api/user/current
12 threads and 400 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 25.38ms 41.24ms 460.39ms 90.51%
Req/Sec 2.38k 325.27 4.46k 88.33%
Latency Distribution
50% 12.09ms
75% 18.72ms
90% 63.93ms
99% 215.51ms
845952 requests in 30.05s, 129.08MB read
Requests/sec: 28151.60
Transfer/sec: 4.30MB
So, this stack in 2.8 times faster than Undertow and Webflux and in 3.18 times faster than Tomcat.
The difference in rates becomes higher with more performance hardware.