rate-limiter

Motivation

When I started this I didn't know that there are a lot of rate-limiting projects(commercial, personal). I just wanted to solve a situation with my code and I learned something while making this project.

Long story short

This is an app that you should put in front of your app to limiting incoming HTTP traffic. Works on L7 and acts as a reverse proxy. I'm sure there are a lot of similar projects but this project is exactly for my need.

Note: Putting in front of means that you should redirect the incoming traffic to this app instead of your app. This app will redirect the traffic to your app if it should.

How to configure

I prefer telling these by using an example configuration. Here is an example configuration.

app:
  port: 8081
  timeout: 3s
  hosts:
    - 12.34.56.78
    - 12.34.56.79
    - 12.34.56.80
server:
  addr: :8080
  readTimeout: 1000
  writeTimeout: 1000
cache:
  inMemory: true
  redis:
    addr: localhost:6379
    username: ""
    password: ""
    db: 0
defaultRuleScope: path
rules:
  - scope: path
    pattern: GET /users/4/addresses
    limit: 4
tracing:
  enabled: false
  provider: new-relic
  newRelic:
    appName: appName
    licenseKey: licenseKey
    distributedTracerEnabled: true

In the app object you fulfill your app's information.

Name	Description	Required	Default
port	Your app port. You can specify this on each of your hosts	yes	None
hosts	Your app instances. Can be host only or host:port	yes	None
timeout	Your app's timeout when redirecting	no	1000ms

In the server object you set up a server.

Name	Description	Required	Default
addr	Address to listen	yes	None
readTimeout	Server read timeout in millisecond format	no	1000ms
writeTimeout	Server read timeout in millisecond format	no	1000ms

In the cache object you specify whether you want to use in-memory cache or Redis. If you use in-memory than the only thing you have to do is set cache.inMemory field as true. However, if you use Redis than you only have to specify the Redis settings.

In the rules array, you specify your rules. The pattern on the rule is basically a Regular Expression. Rules are processed sequentially.

Name	Description	Required	Default
pattern	Regular expression to comparison pattern.	yes	None
limit	Limit of the total requests	yes	0
scope	Specifies the increment key. Can be rule or path. Rule pattern is used as increment key. Otherwise request's METHOD PATH	no	$defaultRuleScope

In the tracing object, you specify a tracing method for this app. We support only NewRelic right now.

Name	Description	Required	Default
enabled	Enabled the tracing	no	false
provider	Your tracing provider. Available providers is: [new-relic]	no	new-relic
newRelic.appName	App name on NewRelic	yes	rate-limiter
newRelic.licenseKey	License key to access to NewRelic	yes	None
newRelic.distributedTracerEnabled	Specifies whether Distributed Tracing is enabled or not	no	false

homeless texts

Your regexp is compared with this format METHOD PATH, for example, you get a request like DELETE /users?name=John but this is compared with your regexp as this format DELETE /users.

Let's make another example. Assume that we have already some rules as defined above. Here is an example curl and the processing of this.

$ curl -X GET localhost:8000/users/4/addresses?country=tr

You can do this request 4 times on each 30 seconds.

Todo

• Implement windowLengthInSeconds
• Kubernetes Operator
• OpenTelemetry Support
• Rich algorithm support