Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time


After investigating all ruby gems and commercial services for testing my API performance, I figured out that none of those was easy to use. Most of them were not built with APIs in mind, were difficult to create a simple test, even more tricky to add more sophisticated requests and paid services didn't work for localhost.

All failed to my request: test and measure my API performance without spending too much time in this shit.

So I built my own gem. Started as a funny gist, ended up as a fully fledged gem.

ApiBomb will allow you to test how much your API can take. It will start firing as many requests as you want for any timespan you want, all of them fully customizable.

Are you ready to defend your API?


Add this line to your application's Gemfile:

gem 'api_bomb'

And then execute:

$ bundle

Or install it yourself as:

$ gem install api_bomb


First you need to define the global settings. Usually in the global settings hash you just want to define the most common settings.

options = {
  concurrent_users: 4, #concurrent users
  duration: 60, #seconds
  base_url: 'http://localhost:3000/api/v1', #base url
  options: { #this hash is overriden for each path that has its own options
    headers: { #various headers, like session token etc
      'Session-Token' => SESSION_TOKEN,
      'Connection' => 'close',
      'Content-Type' => 'application/json'
    params: { #GET params
    json: { #POST/PUT/PATCH HTTP body params

Options keys (like headers, params, json etc) can be overrided later, if needed, per path. Oh btw those options are passed straight to http which means you can do anything this gem provides :)

Simple mode

You want to test how many GET rpm an endpoint can hold?
path = 'videos'{path: path})).start!

This will fire tons of requests in the http://localhost:3000/api/v1/videos using the headers from the global settings for 60 seconds. 4 concurrent users means that they will always be 4 requests pending from the client perspective. As soon as one is served, a new one will be fired. Once the test duration is elapsed it will report back to you:

Elapsed time: 60
Concurrency: 4 threads
Number of requests: 298
Requests per second: 4
Requests per minute: 298.0
Average response time: 0.6628006505031939
Standard deviation: 0.2957473991283155
Percentile 90th: 1.0887710852002783
Percentile 95th: 1.154147314250258
Percentile 99th: 1.5413802972278787
Server status stats: [{"2xx"=>298}]

You can also inject your logger in the global options.

You want to test how many GET rpm a sequence of endpoints can hold? It will run

a separate benchmark for each of your endpoints.

paths = ['videos', 'users', 'comments']{paths: paths})).start!

This will fire tons of requests like in the previous example, first in http://localhost:3000/api/v1/videos then in http://localhost:3000/api/v1/users and finally in http://localhost:3000/api/v1/comments

You want to test a POST endpoint? Then you can the slightly more advanced API:
video_params = { title: 'A new video!', description: 'a new description!', user_id: 1}
paths = {path: 'videos', action: :post, options: { json: video_params}}{paths: paths})).start!
You want to test dynamic endpoints?
paths = {path:{ "videos/#{1.upto(10000).to_a.sample}" } #default action is :get{paths: paths})).start!

You can add lambdas or (preferrably) procs in any value of the options hash and path.

It should be noted that using lambdas/procs, it's almost endless of what you can do. You could use FactoryGirl or whatever to get dynamic attributes when creating/updating a resource in your API. Be sure to watch out though, that constants and classes have already been initialized because most of such gems do lazy initialization.

You want to test a sequence of endpoints with dynamic params? It will run a separate

benchmark for each one in the array.

paths = [ #get http method is used by default
    path: 'videos',
    params: { per_page:{ 1.upto(10).to_a.sample } } #you can have a proc in a specific param only
    path: 'videos',
    params: { per_page:{ 1.upto(10).to_a.sample } } #you can have a proc in a specific param only
    path: 'videos',
    params: { per_page:{ 1.upto(10).to_a.sample } } #you can have a proc in a specific param only
    path: 'videos',
    params: { per_page:{ 1.upto(10).to_a.sample } } #you can have a proc in a specific param only
You want to test dynamic endpoints with dynamic params?
paths = {
  path:{ ['videos', 'users', 'comments', 'likes'].sample,
  action: :get,
  params:{ {per_page: 1.upto(10).to_a.sample} }
}{paths: paths})).start!

This is also equivelent to the previous:

paths = {
  path:{ ['videos', 'users', 'comments', 'likes'].sample,
  action: :get,
  params: { per_page:{ 1.upto(10).to_a.sample } } #you can have a proc in a specific param only
}{paths: paths})).start!
You want to test different endpoints with each having different probability to be requested?

The dynamic nature is up to you. But for your convinience we have created a special class that can be used if you want to test random paths using a probability/weight. So if you want for bombard 'videos' path 3 times more than comments and comments path 3 times more users path (so videos will be bombarded 9 times more than users path) in a war (test):

paths ={
  {path: 'videos'} => 9,
  {path: 'comments'} => 3,
  {path: 'users'} => 1,
}){paths: paths})).start!

or a more spophisticated:

paths ={
  {path: 'videos', action: :get, options: { params: {per_page: 1}}} => 20,
  {path: 'videos', action: :get, options: {
    params: { per_page:{ 1.upto(10).to_a.sample } }
  }} => 10,
  {path:{ "videos/#{video_ids.sample}" }, action: :get} => 70,
  {path:{ "users/#{user_ids.sample}" }, action: :get} => 60,
}){paths: paths})).start!

Weighted path uses Pickup gem underneath.


You can have lambdas/procs in any hash value (or key for Weighted paths). Internally Path::Single, Path::Sequence and Path::Weighted classes are used everywhere, which you can also use but I have ommitted them from the examples for the sake of simplicity (and added a builder that figures out what you actually want).

You can also create your own Path structure. It only needs to respond to pick method which must return a Path::Single object with all the necessary attributes.

Furthermore, if you like, you can override the default strategy. You might want to dispatch a new request depending on the response of the previous request. Take a look on lib/api_bomb/strategies.rb

On the global settings hash, you can also specify the number of requests you want to send (which will override the duration unless Timeout exception kicks in first). It's not very well tested and should be used mostly for debugging (requests: 1)

Best practices

  • When starting optimizing, be sure that you test the performance of your API and not of your webserver. For instance, you might fire 2 concurrent users in an endpoint which will result in 300 req/min. Then you might fire 20 concurrent users which could result in 450 req/min. Does it mean that your API is faster? Probably not because your response time must have went > 1 sec which sucks.

That's why other statistics are included apart from req/min.


  1. Fork it ( )
  2. Create your feature branch (git checkout -b my-new-feature)
  3. Commit your changes (git commit -am 'Add some feature')
  4. Push to the branch (git push origin my-new-feature)
  5. Create a new Pull Request


API load stressing and testing/measuring framework. Are you ready to defend your API ?



Code of conduct





No releases published


No packages published