woo

https://github.com/fukamachi/woo.git

git clone 'https://github.com/fukamachi/woo.git'

(ql:quickload :woo)
★323

Woo

Build Status

Woo is a fast non-blocking HTTP server built on top of libev. Although Woo is written in Common Lisp, it aims to be the fastest web server written in any programming language.

Warning

This library is still under development and considered ALPHA quality.

How fast?

Benchmark graph

See Benchmarks for the detail.

Usage

Start a server

(ql:quickload :woo)

(woo:run
  (lambda (env)
    (declare (ignore env))
    '(200 (:content-type "text/plain") ("Hello, World"))))

Start with Clack

(ql:quickload :clack)

(clack:clackup
  (lambda (env)
    (declare (ignore env))
    '(200 (:content-type "text/plain") ("Hello, World")))
  :server :woo
  :use-default-middlewares nil)

Cluster

(woo:run
  (lambda (env)
    (declare (ignore env))
    '(200 (:content-type "text/plain") ("Hello, World")))
  :worker-num 4)

Benchmarks

Comparison of the server performance to return “Hello, World” for every requests. Here's the results of requests/sec scores.

Benchmark Results

Here's the new graph when using multiple CPU cores:

Benchmark Results (multicore)

All benchmarks were done with the below command of wrk.

wrk -c [10 or 100] -t 4 -d 10 http://127.0.0.1:5000

The benchmarking environment is:

Tornado (Python)

$ python benchmark/hello.py
$ wrk -c 10 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     2.83ms  218.41us   6.63ms   96.11%
    Req/Sec   735.93     83.69     0.89k    80.07%
  28223 requests in 10.00s, 5.57MB read
Requests/sec:   2822.09
Transfer/sec:    570.48KB
$ wrk -c 100 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency    39.14ms    1.02ms  45.78ms   73.21%
    Req/Sec   641.91     37.37   825.00     82.48%
  25504 requests in 10.01s, 5.03MB read
Requests/sec:   2548.72
Transfer/sec:    515.22KB

Hunchentoot (Common Lisp)

$ sbcl --load benchmark/hunchentoot.lisp
$ wrk -c 10 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   144.00us   97.95us   4.76ms   99.85%
    Req/Sec     6.81k   509.00     7.22k    97.05%
  63978 requests in 10.01s, 9.58MB read
  Socket errors: connect 0, read 0, write 0, timeout 34
Requests/sec:   6390.08
Transfer/sec:      0.96MB
wrk -c 100 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     3.24ms    2.45ms   5.19ms   61.52%
    Req/Sec     6.83k   556.53     7.89k    93.36%
  64140 requests in 10.01s, 9.60MB read
  Socket errors: connect 0, read 161, write 0, timeout 387
Requests/sec:   6406.92
Transfer/sec:      0.96MB

Tornado (PyPy)

$ pypy benchmark/hello.py
$ wrk -c 10 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     4.21ms    8.64ms  30.26ms   90.41%
    Req/Sec     1.80k     1.11k    3.67k    56.02%
  67132 requests in 10.00s, 13.25MB read
Requests/sec:   6712.27
Transfer/sec:      1.33MB
$ wrk -c 100 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency    16.08ms    8.16ms  40.17ms   89.97%
    Req/Sec     1.60k   311.97     2.16k    63.06%
  62962 requests in 10.01s, 12.43MB read
Requests/sec:   6292.83
Transfer/sec:      1.24MB

Wookie (Common Lisp)

$ sbcl --load benchmark/wookie.lisp
$ wrk -c 10 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.25ms    3.35ms  52.53ms   99.16%
    Req/Sec     1.99k   227.28     2.33k    78.41%
  75538 requests in 10.00s, 8.28MB read
Requests/sec:   7554.20
Transfer/sec:    848.37KB
$ wrk -c 100 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency    13.15ms    1.89ms  25.52ms   85.12%
    Req/Sec     1.94k   349.30     2.59k    64.35%
  76227 requests in 10.01s, 8.36MB read
Requests/sec:   7618.58
Transfer/sec:    855.60KB

Node.js http module

$ node benchmark/node.js
$ wrk -c 10 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   558.17us   57.01us   4.02ms   95.83%
    Req/Sec     3.73k   213.97     4.11k    63.56%
  141101 requests in 10.00s, 17.49MB read
Requests/sec:  14110.16
Transfer/sec:      1.75MB
$ wrk -c 100 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     7.71ms    0.91ms  26.78ms   93.24%
    Req/Sec     3.37k   366.65     4.53k    74.85%
  130057 requests in 10.00s, 16.12MB read
Requests/sec:  13003.58
Transfer/sec:      1.61MB

Woo (Common Lisp)

$ sbcl --load benchmark/woo.lisp
$ wrk -c 10 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   304.78us  435.89us  10.34ms   99.56%
    Req/Sec     7.06k     1.33k    9.89k    74.84%
  267269 requests in 10.00s, 33.65MB read
Requests/sec:  26728.29
Transfer/sec:      3.36MB
$ wrk -c 100 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     3.83ms    1.35ms  23.13ms   95.43%
    Req/Sec     7.02k     1.03k    8.33k    82.96%
  266506 requests in 10.00s, 33.55MB read
Requests/sec:  26649.52
Transfer/sec:      3.35MB

Go

$ go build benchmark/hello.go
$ ./hello
$ wrk -c 10 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   274.14us  142.62us   0.92ms   57.60%
    Req/Sec     7.47k     0.96k   10.60k    64.57%
  282965 requests in 10.00s, 34.81MB read
Requests/sec:  28298.26
Transfer/sec:      3.48MB
wrk -c 100 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     3.72ms  742.47us   6.22ms   62.44%
    Req/Sec     7.09k   568.00     9.00k    65.49%
  269210 requests in 10.00s, 33.12MB read
Requests/sec:  26922.81
Transfer/sec:      3.31MB

Unicorn + nginx (Ruby, worker_processes=4)

nginx's worker_processes=4 Unicorn's worker_processes=4

$ unicorn -E production -c benchmark/unicorn/config.rb benchmark/unicorn/app.ru
$ wrk -c 10 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.56ms    9.56ms 108.21ms   98.77%
    Req/Sec     4.09k     1.40k    9.22k    74.49%
  153585 requests in 10.00s, 24.45MB read
Requests/sec:  15361.96
Transfer/sec:      2.45MB
$ wrk -c 100 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     6.60ms    4.55ms 137.92ms   89.16%
    Req/Sec     3.96k   657.47    11.31k    81.94%
  152428 requests in 10.00s, 24.27MB read
Requests/sec:  15243.51
Transfer/sec:      2.43MB

Node.js http module (4 cluster)

$ node benchmark/node-cluster.js
$ wrk -c 10 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   334.44us  570.21us  19.06ms   90.08%
    Req/Sec     7.46k     5.29k   20.44k    63.69%
  280336 requests in 10.00s, 34.76MB read
Requests/sec:  28035.25
Transfer/sec:      3.48MB
$ wrk -c 100 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     3.85ms    5.98ms 108.27ms   95.20%
    Req/Sec     7.73k     2.65k   31.56k    74.38%
  285013 requests in 10.00s, 35.34MB read
Requests/sec:  28498.97
Transfer/sec:      3.53MB

Woo (Common Lisp, worker-num=4)

$ sbcl --load benchmark/woo-cluster.lisp
$ wrk -c 10 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.08ms   12.66ms 184.39ms   99.53%
    Req/Sec     9.86k     2.25k   15.00k    68.51%
  373207 requests in 10.00s, 46.98MB read
Requests/sec:  37323.22
Transfer/sec:      4.70MB
$ wrk -c 100 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     2.86ms    9.19ms 126.45ms   96.01%
    Req/Sec    13.39k     6.71k   45.78k    67.70%
  504881 requests in 10.00s, 63.56MB read
Requests/sec:  50505.20
Transfer/sec:      6.36MB

Go (GOMAXPROCS=4)

$ go build benchmark/hello-maxproc.go
$ ./hello-maxproc
$ wrk -c 10 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   141.53us  140.85us   8.08ms   95.00%
    Req/Sec    13.50k     3.15k   24.44k    71.60%
  510975 requests in 10.00s, 62.86MB read
Requests/sec:  51101.54
Transfer/sec:      6.29MB
$ wrk -c 100 -t 4 -d 10 http://127.0.0.1:5000
Running 10s test @ http://127.0.0.1:5000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.75ms    1.58ms  35.92ms   95.40%
    Req/Sec    15.55k     5.36k   62.67k    82.92%
  583243 requests in 10.00s, 71.75MB read
Requests/sec:  58336.29
Transfer/sec:      7.18MB

Phusion Passenger (Ruby)

$ passenger start -e production -R benchmark/unicorn/app.ru --max-pool-size 4 --min-instances 4
wrk -c 10 -t 4 -d 10 http://127.0.0.1:3000
Running 10s test @ http://127.0.0.1:3000
  4 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency   177.00us    1.10ms  45.88ms   99.64%
    Req/Sec    13.87k     3.06k   27.00k    80.29%
  526936 requests in 10.00s, 79.90MB read
Requests/sec:  52696.83
Transfer/sec:      7.99MB
wrk -c 100 -t 4 -d 10 http://127.0.0.1:3000
Running 10s test @ http://127.0.0.1:3000
  4 threads and 100 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.42ms    1.85ms  64.67ms   98.36%
    Req/Sec    15.60k     3.55k   47.00k    62.83%
  590815 requests in 10.00s, 89.59MB read
  Socket errors: connect 0, read 0, write 0, timeout 48
Requests/sec:  59085.92
Transfer/sec:      8.96MB

Installation

Woo has switched the backend from cl-async to libev after the latest Quicklisp dist release. If you're gonna run the benchmarks by your own, please use the latest one.

Requirements

Cloning

$ cd ~/common-lisp
$ git clone https://github.com/fukamachi/lev
$ git clone https://github.com/fukamachi/woo

Loading

(ql:quickload :woo)

TODO

See Also

Author

Copyright

Copyright (c) 2014 Eitaro Fukamachi (e.arrows@gmail.com)

License

Licensed under the MIT License.