Skip to content

Kaligo/heavy_keeper

Repository files navigation

HeavyKeeper

This gem implements HeavyKeeper algorithm, based on the paper with the same name: https://www.usenix.org/conference/atc18/presentation/gong

The interface is designed to be similar with ReBloom TopK datastructure (DS) (https://oss.redis.com/redisbloom/TopK_Commands/).

This is a naive implementation of HeavyKeeper, probably not very optimized. We use multiple Redis DSs:

  • A hash with maximum depth * width items to act as a bucket to store main counter
  • A sorted set with maximum K elements to act as a MinHeap
  • A hash with maximum K elements to store more correct counter of the element in MinHeap

Installation

Add this line to your application's Gemfile:

gem 'heavy_keeper'

Usage

You will need to add an initializer to provide some configuration:

HeavyKeeper.configure do |config|
  config.cache_prefix = 'cache_prefix'.freeze # currently used as prefix for the redis data structures.
  config.storage = Redis.new # a Redis store, at least version 4.0
end

In general, you will only interact with an instance of the HeavyKeeper::TopK class.

These are the most relevant instance methods (see code comments for more detail):

reserve(name, top_k: size, width:, depth:, decay:) - sets up a Top K list with specified options

increase_by(name, *items) - add an array of items to a list

list(name) - returns full list of items in Top K list

clear(name) - deletes list

remove(name, item) - reset the counter of the targeted item in the list

Development

After checking out the repo, run bin/setup to install dependencies. Then, run bundle exec rspec to run the tests.

To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and the created tag, and push the .gem file to rubygems.org.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/Kaligo/heavy_keeper.

About

Gem which implements the HeavyKeeper algorithm

Resources

Stars

Watchers

Forks

Packages

No packages published