README.rst

ATTENTION Boom is going to be replaced by Salvo and then deprecated

Follow the progress and join the party here: https://github.com/tarekziade/salvo

ATTENTION

Boom! is a simple command line tool to send some load to a web app.

Boom! is a script you can use to quickly smoke-test your web app deployment. If you need a more complex tool, I'd suggest looking at Funkload or Locust.

Boom! was specifically written to replace my Apache Bench usage, to provide a few missing features and fix a few annoyances I had with AB.

I have no special ambitions for this tool, and since I have not found any tool like this in the Python-land, I wrote this one.

There are a lot of other tools out there, like Siege which seems very popular.

However, Boom! is a good choice because it works on any platform and is able to simulate thousands of users by using greenlets.

Installation

Boom! requires Gevent and Requests. If you are under Windows I strongly recommend installing Gevent with the xxx-win32-py2.7.exe installer you will find at: https://github.com/surfly/gevent/downloads

Boom! should work with the latest versions.

If you are under Linux, installing the source version is usually a better idea. You will need libev for Gevent.

Example under Ubuntu:

$ sudo apt-get install libev libev-dev python-dev

Then:

$ pip install boom

Basic usage

Basic usage example: 100 queries with a maximum concurrency of 10 users:

$ boom http://localhost:80 -c 10 -n 100
Server Software: nginx/1.2.2
Running 100 queries - concurrency: 10.
Starting the load [===================================] Done

-------- Results --------
Successful calls        100
Total time              0.3260 s
Average                 0.0192 s
Fastest                 0.0094 s
Slowest                 0.0285 s
Amplitude               0.0191 s
RPS                     306
BSI                     Pretty good

-------- Legend --------
RPS: Request Per Second
BSI: Boom Speed Index

Boom! has more options:

$ boom --help
usage: boom [-h] [--version] [-m {GET,POST,DELETE,PUT,HEAD,OPTIONS}]
            [--content-type CONTENT_TYPE] [-D DATA] [-c CONCURRENCY] [-a AUTH]
            [--header HEADER] [--pre-hook PRE_HOOK] [--post-hook POST_HOOK]
            [--json-output] [-n REQUESTS | -d DURATION]
            [url]

Simple HTTP Load runner.

positional arguments:
  url                   URL to hit

optional arguments:
  -h, --help            show this help message and exit
  --version             Displays version and exits.
  -m {GET,POST,DELETE,PUT,HEAD,OPTIONS}, --method {GET,POST,DELETE,PUT,HEAD,OPTIONS}
                        HTTP Method
  --content-type CONTENT_TYPE
                        Content-Type
  -D DATA, --data DATA  Data. Prefixed by "py:" to point a python callable.
  -c CONCURRENCY, --concurrency CONCURRENCY
                        Concurrency
  -a AUTH, --auth AUTH  Basic authentication user:password
  --header HEADER       Custom header. name:value
  --pre-hook PRE_HOOK   Python module path (eg: mymodule.pre_hook) to a
                        callable which will be executed before doing a request
                        for example: pre_hook(method, url, options). It must
                        return a tupple of parameters given in function
                        definition
  --post-hook POST_HOOK
                        Python module path (eg: mymodule.post_hook) to a
                        callable which will be executed after a request is
                        done for example: eg. post_hook(response). It must
                        return a given response parameter or raise an
                        `boom.boom.RequestException` for failed request.
  --json-output         Prints the results in JSON instead of the default
                        format
  -n REQUESTS, --requests REQUESTS
                        Number of requests
  -d DURATION, --duration DURATION
                        Duration in seconds

Calling from Python code

You can trigger load testing from Python code by importing the function boom.boom.load directly, as follows:

from boom.boom import load
result = load('http://example.com/', 1, 1, 0, 'GET', None, 'text/plain', None, quiet=True)

Design

Boom uses greenlets through Gevent to create virtual users, and uses Requests to do the queries.

Using greenlets allows Boom to spawn large amounts of virtual users with very little resources. It's not a problem to spawn 1000 users and hammer a web application with them.

If you are interested in this project, you are welcome to join the fun at https://github.com/tarekziade/boom

Make sure to add yourself to the contributors list if your PR gets merged. And make sure it's in alphabetical order!