Rate Limiting

Rate limit how many HTTP requests a developer can make in a given period of seconds, minutes, hours, days, months or years. If the underlying Service/Route (or deprecated API entity) has no authentication layer, the Client IP address will be used, otherwise the Consumer will be used if an authentication plugin has been configured.

Terminology

• plugin: a plugin executing actions inside Kong before or after a request has been proxied to the upstream API.
• Service: the Kong entity representing an external upstream API or microservice.
• Route: the Kong entity representing a way to map downstream requests to upstream services.
• Consumer: the Kong entity representing a developer or machine using the API. When using Kong, a Consumer only communicates with Kong which proxies every call to the said upstream API.
• Credential: a unique string associated with a Consumer, also referred to as an API key.
• upstream service: this refers to your own API/service sitting behind Kong, to which client requests are forwarded.

Configuration

This plugin is compatible with requests with the following protocols:

• http
• https

This plugin is partially compatible with DB-less mode.

Enabling the plugin on a Service

With a database

Configure this plugin on a Service by making the following request:

$ curl -X POST http://kong:8001/services/{service}/plugins \
    --data "name=rate-limiting"  \
    --data "config.second=5" \
    --data "config.hour=10000"

Without a database

Configure this plugin on a Service by adding this section do your declarative configuration file:

plugins:
- name: rate-limiting
  service: {service}
  config: 
    second: 5
    hour: 10000

In both cases, {service} is the id or name of the Service that this plugin configuration will target.

Enabling the plugin on a Route

With a database

Configure this plugin on a Route with:

$ curl -X POST http://kong:8001/routes/{route}/plugins \
    --data "name=rate-limiting"  \
    --data "config.second=5" \
    --data "config.hour=10000"

Without a database

Configure this plugin on a Route by adding this section do your declarative configuration file:

plugins:
- name: rate-limiting
  route: {route}
  config: 
    second: 5
    hour: 10000

In both cases, {route} is the id or name of the Route that this plugin configuration will target.

Enabling the plugin on a Consumer

With a database

You can use the http://localhost:8001/plugins endpoint to enable this plugin on specific Consumers:

$ curl -X POST http://kong:8001/consumers/{consumer}/plugins \
    --data "name=rate-limiting" \
     \
    --data "config.second=5" \
    --data "config.hour=10000"

Without a database

Configure this plugin on a Consumer by adding this section do your declarative configuration file:

plugins:
- name: rate-limiting
  consumer: {consumer}
  config: 
    second: 5
    hour: 10000

In both cases, {consumer} is the id or username of the Consumer that this plugin configuration will target.

You can combine consumer_id and service_id

in the same request, to furthermore narrow the scope of the plugin.

Global plugins

• Using a database, all plugins can be configured using the http://kong:8001/plugins/ endpoint.
• Without a database, all plugins can be configured via the plugins: entry on the declarative configuration file.

A plugin which is not associated to any Service, Route or Consumer (or API, if you are using an older version of Kong) is considered "global", and will be run on every request. Read the Plugin Reference and the Plugin Precedence sections for more information.

Parameters

Here's a list of all the parameters which can be used in this plugin's configuration:

Headers sent to the client

When this plugin is enabled, Kong will send some additional headers back to the client telling how many requests are available and what are the limits allowed, for example:

X-RateLimit-Limit-Minute: 10
X-RateLimit-Remaining-Minute: 9

or it will return a combination of more time limits, if more than one is being set:

X-RateLimit-Limit-Second: 5
X-RateLimit-Remaining-Second: 4
X-RateLimit-Limit-Minute: 10
X-RateLimit-Remaining-Minute: 9

If any of the limits configured is being reached, the plugin will return a HTTP/1.1 429 status code to the client with the following JSON body:

{"message":"API rate limit exceeded"}

Implementation considerations

The plugin supports 3 policies, which each have their specific pros and cons.

There are 2 use cases that are most common:

1. every transaction counts. These are for example transactions with financial consequences. Here the highest level of accuracy is required.
2. backend protection. This is where accuracy is not as relevant, but it is merely used to protect backend services from overload. Either by specific users, or to protect against an attack in general.

NOTE:

Every transaction counts

In this scenario, the local policy is not an option. So here the decision is between the extra performance of the redis policy against its extra support effort. Based on that balance, the choice should either be cluster or redis.

The recommendation is to start with the cluster policy, with the option to move over to redis if performance reduces drastically. Keep in mind existing usage metrics cannot be ported from the datastore to redis. Generally with shortlived metrics (per second or per minute) this is not an issue, but with longer lived ones (months) it might be, so you might want to plan your switch more carefully.

Backend protection

As accuracy is of lesser importance, the local policy can be used. It might require some experimenting to get the proper setting. For example, if the user is bound to 100 requests per second, and you have an equally balanced 5 node Kong cluster, setting the local limit to something like 30 requests per second should work. If you are worried about too many false-negatives, increase the value.

Keep in mind as the cluster scales to more nodes, the users will get more requests granted, and likewise when the cluster scales down the probability of false-negatives increases. So in general, update your limits when scaling.

The above mentioned inaccuracy can be mitigated by using a consistent-hashing load balancer in front of Kong, that ensures the same user is always directed to the same Kong node. This will both reduce the inaccuracy and prevent the scaling issues.

Most likely the user will be granted more than was agreed when using the local policy, but it will effectively block any attacks while maintaining the best performance.