Sliding Window Counter Rate Limiter

Python sliding window counter API rate limiter using Redis.

Table of contents

• Why Sliding window counter
• Requirements
• Rate limit strategies
• Rate limit parameters
• Usage
• Rate limiter supported methods
• Optional parameters
• License
• Contributing

Why Sliding window counter

Sliding window counter is used for two main reasons which are

• Avoid spikes (smooth out bursts)
• Memory efficient

Requirements

• Python
• Redis

Rate limit strategies

These strategies are used in rate limiter.

• N requests per second
• N requests per minute
• N requests per hour

Rate limit parameters

These parameter must be passed when initialize rate limiter class.

• clientid
• redispipeline
• rate
• time_window_unit

This clientid parameter is used for how many requests serverd for particular user and how many request are left.

This redispipeline parameter is redis pipeline which is used for perform opeartions in single shot.

This rate parameter is used for how many request is allowed for particular user.

rate = 100 # 100 request are allowed  

This time_window_unit parameter is used for time window.

# singular time unit allowed. For example 

time_window_unit = 'hour'  
time_window_unit = 'minute'
time_window_unit = 'second'

Both rate and time_window_unit parameters are used for rate limit. Set rate limit

1400 requests per hour.

rate = 1400
time_window_unit = 'hour'

150 requests per minute.

rate = 150
time_window_unit = 'minute'

12 requests per second.

rate = 12
time_window_unit = 'second'

Usage

Import redis and sliding window counter class

import redis
from  ratelimiter import SlidingWindowCounterRateLimiter

Create redis pipeline

r = redis.Redis(host='localhost', port=6379, db=1)
pipeline = r.pipeline()

We need hourly rate limit for example 300 requests per hour.So initialize rate limiter

rate = 300 # 300 requests allowed
time_window_unit = 'hour' # per hour
client_id = 'user-100C' # client id 

# pass these argument into rate limiter
ratelimiter = SlidingWindowCounterRateLimiter(clientid=client_id,redispipeline=pipeline,rate = rate,time_window_unit=time_window_unit)

After that check request is allowed or not.

if ratelimiter.isRequestAllowed(): # Return true if request allowed 
    print('200')

else: # return false if request not allowed
    print('429')

Rate limiter supported methods

Get rate limit HTTP headers.These HTTP headers are returns as a dictionary.

• X-RateLimit-Limit
• X_RateLimit_Remaining
• X-RateLimit-Reset : This header used extra space so if this header set true then it will be part of rate limit http headers otherwise not.
• Retry-after

ratelimiter.getHttpResponseHeaders() # return HTTP response headers as a dictionary.

Get Retry-after time.

ratelimiter.get_retry_after()

Get X_RateLimit_Remaining.

ratelimiter.get_x_ratelimit_remaining()

Get total request served

ratelimiter.getTotalRequestServedInSlidingWindow()

Get rate limit (max request allowed)

ratelimiter.getMaxRequestsAllowed() 

Optional parameters

Optional parameters can be passed to rate limiter when first time rate limiter initialized. These parameters are used for performance and reduce memory usage.

This is_ratelimit_reset_header_allowed parameter is used for X-RateLimit-Reset header which used extra space. If not required then set to false when initialize ratelimiter.

is_ratelimit_reset_header_allowed = True # By default set to true

This is_2nd_RTT_allowed parameter is used for reduce memory usage and but increase latency. Hhen set to true on each request it will delete any expired Fixed time window which saved memory but increase latency.

is_2nd_RTT_allowed = False  # RTT = Round Trip Time

This max_no_time_window_for_deletion parameter is used for both performance and reduce memory usage. When expired Fixed time window reached/euqal to this parameter then it will delete all expired Fixed time windows and it will ignore is_2nd_RTT_allowed parameter if set to true of false.

max_no_time_window_for_deletion = 5  # By default set to 5 

License

License under a MIT License.

Contributing

• Fork, clone or make a pull requset to this repository.
• Ask here https://github.com/faizanfareed/python-sliding-window-counter-rate-limiter/issues