Hackage :: [Package]

uhttpc: Minimal HTTP client library optimized for benchmarking

This is a package candidate release! Here you can preview how this package release will appear once published to the main package index (which can be accomplished via the 'maintain' link below). Please note that once a package has been published to the main package index it cannot be undone! Please consult the package uploading documentation for more information.

[maintain] [Publish]

uhttpc contains a simple low-level and lightweight Haskell HTTP 1.1 library optimized for providing the bare minimum required for HTTP benchmarking purposes and allowing for time measurements of the individual phases of the HTTP transaction.

This package also provides the command-line tool uhttpc-bench as an usage example which mimics the popular weighttp HTTP benchmarking tool's CLI.

WARNING: uhttpc was originally developed for evaluating the new "Mio" parallel I/O manager introduced with GHC 7.8 and provide a base-line for HTTP client libraries. However, this is not a fully RFC compliant HTTP client library and therefore ought not be used as a general purpose HTTP implementation; Use libraries such as http-streams instead which aim toward full RFC compliance as well as having good performance.

[Skip to Readme]

Properties

Versions 0.1.0.0, 0.1.1.0, 0.1.1.1, 0.1.1.5
Change log CHANGELOG.md
Dependencies async (>=2.0 && <2.2), base (>=4.6 && <4.11), bytestring (>=0.10 && <0.11), bytestring-lexing (>=0.4 && <0.6), deepseq (>=1.1 && <1.5), network (>=2.6 && <2.7), network-uri (>=2.6 && <2.7), optparse-applicative (>=0.11 && <0.14), uhttpc [details]
License GPL-3.0-only
Copyright © 2013-2015 Herbert Valerio Riedel
Author Herbert Valerio Riedel
Maintainer hvr@gnu.org
Category Network
Home page https://github.com/hvr/uhttpc
Source repo head: git clone https://github.com/hvr/uhttpc.git
Uploaded by HerbertValerioRiedel at 2018-01-05T15:15:33Z

Modules

Downloads

Maintainer's Corner

Package maintainers

For package maintainers and hackage trustees

Readme for uhttpc-0.1.1.5

[back to package description]

uhttpc - µHTTP client library

uhttpc is a simple low-level and lightweight Haskell HTTP 1.1 library providing the bare minimum required for HTTP benchmarking purposes.

This is not a RFC compliant HTTP client library and shall not be used as a general purpose HTTP implementation!

uhttpc-bench

This Cabal package comes with an executable uhttpc-bench which represents an ab/weighttpd-style HTTP benchmarking tool:

uhttpc-bench - a Haskell-based ab/weighttp-style webserver benchmarking tool

Simple HTTP benchmark tool similiar to ab and weighttp

uttpc-bench [OPTIONS] <url>

Common flags:
  -n=num                     number of requests    (default: 1)
  -t=num                     threadcount           (default: 1)
  -c=num                     concurrent clients    (default: 1)
  -k                         enable keep alive
          --csv=FILE         dump request timings as CSV (RFC4180) file
          --user-agent=ITEM  specify User-Agent    (default: "httpc-bench")
  -H=str                     add header to request
  -v      --verbose          enable more verbose statistics and output
          --no-stats         disable statistics
  -p=FILE                    perform POST request with file-content as body
  -l=FILE                    perform a POST request per line, no quoting,
                             round-robin, each client independently
  -?      --help             Display help message
  -V      --version          Print version information

How to use it

First, install the uhttpc package (This requires GHC 7.6.x or later)

$ cabal install uhttpc

A simple example invocation:

$ uhttpc-bench -n 100000 -t1 -c3 -k http://localhost/

uhttpc-bench - a Haskell-based ab/weighttp-style webserver benchmarking tool

starting benchmark...
finished in 2.314289 seconds, 100000 reqs (3 conns), 43209.8 req/s received
status codes: 100000 HTTP-200
data received: 36205.098 KiB/s, 85800000 bytes total (24600000 bytes http + 61200000 bytes content)
rtt min/avg/max = 0.034/0.068/4.252 ms

For comparison, weighttp can be invoked with the very same arguments (in this case at least):

$ weighttp  -n 100000 -t1 -c3 -k http://localhost/

weighttp - a lightweight and simple webserver benchmarking tool

starting benchmark...
spawning thread #1: 3 concurrent requests, 100000 total requests

finished in 2 sec, 333 millisec and 421 microsec, 42855 req/s, 35908 kbyte/s
requests: 100000 total, 100000 started, 100000 done, 100000 succeeded, 0 failed, 0 errored
status codes: 100000 2xx, 0 3xx, 0 4xx, 0 5xx
traffic: 85800000 bytes total, 24600000 bytes http, 61200000 bytes data

Another example invocation of uhttpc-bench:

$ uhttpc-bench http://www.google.com/ -v -c8 -n1000

uhttpc-bench - a Haskell-based ab/weighttp-style webserver benchmarking tool

using 88-byte request header (+ 0-byte body):
 "GET / HTTP/1.1\r\nHost: www.google.com:80\r\nUser-Agent: uhttpc-bench\r\nConnection: close\r\n\r\n"

starting benchmark...

per-client stats:

 client spawned +0.000008 s, 125 reqs (125 conns), 8.1 req/s, finished in 15.369619 s
 rtt min/avg/med/max = 99.586/122.902/122.109/167.193 ms

 client spawned +0.000258 s, 125 reqs (125 conns), 8.1 req/s, finished in 15.383278 s
 rtt min/avg/med/max = 100.494/123.012/121.648/184.555 ms

 client spawned +0.000292 s, 125 reqs (125 conns), 8.1 req/s, finished in 15.365843 s
 rtt min/avg/med/max = 98.721/122.872/121.668/163.810 ms

 client spawned +0.000327 s, 124 reqs (124 conns), 8.1 req/s, finished in 15.346788 s
 rtt min/avg/med/max = 103.008/123.709/122.834/174.003 ms

 client spawned +0.000366 s, 126 reqs (126 conns), 8.2 req/s, finished in 15.367205 s
 rtt min/avg/med/max = 95.255/121.907/120.736/152.943 ms

 client spawned +0.000403 s, 124 reqs (124 conns), 8.1 req/s, finished in 15.357957 s
 rtt min/avg/med/max = 97.730/123.800/123.569/162.326 ms

 client spawned +0.000434 s, 125 reqs (125 conns), 8.1 req/s, finished in 15.388717 s
 rtt min/avg/med/max = 103.056/123.055/121.961/162.419 ms

 client spawned +0.000461 s, 126 reqs (126 conns), 8.2 req/s, finished in 15.394365 s
 rtt min/avg/med/max = 102.114/122.123/121.289/151.520 ms

finished in 15.394867 seconds, 1000 reqs (1000 conns), 65.0 req/s received
status codes: 1000 HTTP-302
data received: 63.498 KiB/s, 1001000 bytes total (783000 bytes http + 218000 bytes content)
rtt 2/9|25/50/75|91/98-th quantile = 103.737/108.696 | 115.124/121.688/129.944 | 137.715/148.159 ms
rtt min/avg/max = 95.255/122.919/184.555 ms

The --csv option allows to export the raw measurement data in format suitable for offline analysis with statistical tools such as R.