Benchmark HTTP proxies

docker alpine 4 instances with pm2 - 7869reqs

unitech@t450: ~/keymetrics/rocky-lb
>>> wrk -c 100 -t 100 -d 10 http://localhost:3051
Running 10s test @ http://localhost:3051
  100 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency    31.79ms  137.82ms   1.89s    97.08%
    Req/Sec    97.57     59.72   470.00     80.69%
  79474 requests in 10.10s, 9.17MB read
  Socket errors: connect 0, read 0, write 0, timeout 5
Requests/sec:   7869.75
Transfer/sec:      0.91MB

http.js

var http = require('http');

var server = http.createServer(function(req, res) {
  res.writeHead(200);
  res.end('hey');
}).listen(process.env.PORT || 8001, function() {
  console.log('App listening on port 8001');
});

nginx upstream 8445.57 reqs

unitech@t450: ~/keymetrics/rocky-lb
>>> wrk -c 100 -t 100 -d 10 http://localhost:3003
Running 10s test @ http://localhost:3003
  100 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency    11.96ms    4.92ms  47.79ms   86.88%
    Req/Sec    85.26     27.17   505.00     84.21%
  85303 requests in 10.10s, 12.36MB read
Requests/sec:   8445.57
Transfer/sec:      1.22MB

pm2 cluster mode -i 1 20220 reqs

unitech@t450: ~/keymetrics/rocky-lb
>>> wrk -c 100 -t 100 -d 10 http://localhost:8001
Running 10s test @ http://localhost:8001
  100 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     5.95ms   12.66ms 315.73ms   98.89%
    Req/Sec   204.89     28.93   570.00     91.32%
  204231 requests in 10.10s, 23.57MB read
Requests/sec:  20220.96
Transfer/sec:      2.33MB

pm2 cluster mode -i 2 34382reqs

unitech@t450: ~/keymetrics/rocky-lb
>>> wrk -c 100 -t 100 -d 10 http://localhost:8001
Running 10s test @ http://localhost:8001
  100 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     2.98ms    1.76ms  76.16ms   97.04%
    Req/Sec   346.33     58.89     1.91k    79.01%
  347233 requests in 10.10s, 40.07MB read
Requests/sec:  34382.30
Transfer/sec:      3.97MB

raw wrk 24453 reqs

unitech@t450: ~/rocky-lb/wrk
>>> wrk -c 100 -t 100 -d 10 http://localhost:8001
Running 10s test @ http://localhost:8001
  100 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     4.11ms    1.06ms  30.05ms   96.07%
    Req/Sec   246.53     25.70   494.00     96.38%
  246977 requests in 10.10s, 28.50MB read
Requests/sec:  24453.03
Transfer/sec:      2.82MB

rocky lb 1484reqs

unitech@t450: ~/rocky-lb/wrk
>>> wrk -c 100 -t 100 -d 10 http://localhost:3000
Running 10s test @ http://localhost:3000
  100 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency    66.87ms   13.40ms 179.84ms   77.24%
    Req/Sec    15.03      5.13    30.00     98.57%
  14997 requests in 10.10s, 1.87MB read
Requests/sec:   1484.76
Transfer/sec:    189.95KB

tcp-proxy 21889reqs

unitech@t450: ~/keymetrics/rocky-lb
>>> wrk -c 100 -t 100 -d 10 http://localhost:8000
Running 10s test @ http://localhost:8000
  100 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     4.62ms    1.80ms  44.29ms   88.20%
    Req/Sec   220.71     30.77   474.00     84.32%
  221086 requests in 10.10s, 25.51MB read
Requests/sec:  21889.59
Transfer/sec:      2.53MB

test with raw lb nodejs stream
https://github.com/observing/haproxy