RedisMemo

# -- Gemfile --
gem 'redis-memo'

⚠️ Required Redis Version: >= 6.0.0 (reference)

In the User model:

class User < ApplicationRecord
  extend RedisMemo::MemoizeQuery
  memoize_table_column :id
end

SELECT "users".* FROM "users" WHERE "users"."id" = $1 queries will load the data from Redis instead of the database:

[1] (rails console)> Post.last.author
  Post Load (0.5ms)  SELECT "posts".* FROM "posts" ORDER BY "posts"."id" DESC LIMIT $1
[Redis] User Load SELECT "users".* FROM "users" WHERE "users"."id" = $1 LIMIT $2
[Redis] command=MGET args="RedisMemo::Memoizable:wBHc40/aONKsqhl6C51RyF2RhRM=" "RedisMemo::Memoizable:fMu973somRtsGSPlWfQjq0F8yh0=" "RedisMemo::Memoizable:xjlaWFZ6PPfdd8hCQ2OjJi6i0hw="
[Redis] call_time=0.54 ms
[Redis] command=MGET args="RedisMemo:SELECT \"users\".* FROM \"users\" WHERE \"users\".\"id\" = ? LIMIT ?::P+HaeUnujDi9eH7jZfkTzWuv6CA="
[Redis] call_time=0.44 ms
=> #<User id: 1>

Learn more here.

Some computation might depend on multiple database records, for example:

class Post < ApplicationRecord
  extend RedisMemo::MemoizeMethod
  def display_title
    "#{title} by #{author.display_name}"
  end
  memoize_method :display_title do |post|
    depends_on Post.where(id: post.id)
    depends_on User.where(id: post.author_id)
  end
end

• Note that calling Post.where(id: post.id) does not trigger any database queries -- it's just an ActiveRecord Relation representing the SQL query.

In order to use depends_on to extract dependencies from a Relation, we need to memoize the referenced table columns on the Post and User model:

class Post < ApplicationRecord
  extend RedisMemo::MemoizeQuery
  memoize_table_column :id
end

It's also possible to pull in existing dependencies on other memoized methods and perform hierarchical caching.

When a method does not have any dependencies other than its arguments, it is considered a pure function. Pure functions can be cached on Redis as follow:

class FibonacciSequence
  extend RedisMemo::MemoizeMethod

  def [](i); i <= 2 ? 1 : self[i - 1] + self[i - 2]; end
  memoize_method :[]
end

The method arguments are used as part of the cache key to store the actual computation result on Redis.

When a method’s result can not only be derived from its arguments, set dependencies explicitly as follow:

• Call invalidate in after_save
• Set dependencies in memoize_method

class Document
  extend RedisMemo::MemoizeMethod

  def memoizable
    @memoizable ||= RedisMemo::Memoizable.new(document_id: id)
  end

  def after_save
     RedisMemo::Memoizable.invalidate([memoizable])
  end

  # Make an API request to load the document, for example, from AWS S3
  def load; end

  memoize_method :load do |doc|
    depends_on doc.memoizable
  end
end

For each load call, the cached result on Redis will be used until its dependencies have been invalidated.

You can configure various RedisMemo options in your initializer config/initializers/redis_memo.rb:

RedisMemo.configure do |config|
  config.expires_in = 3.hours
  config.global_cache_key_version = SecureRandom.uuid
  ...
end

Learn more here.

1. Database caching: Quick review of why caching is important

2. Challenges with application-level caching 2.1 Forgetting to invalidate the cache 2.2 Cache invalidation during database transactions 2.3 Cache invalidation could be slow and expensive 2.4 Possible race conditions 2.5 Cache inconsistency during deployments

3. How caching is easily done with RedisMemo 3.1 Performant and reliable cache invalidation 3.2 Auto-invalidation 3.3 Add caching without changing any call sites 3.4 Add caching confidently     3.4.1 Avoid mistakes by pulling in existing dependencies     3.4.2 Monitoring     3.4.3 Safely roll out changes

We’re aware of Shopify/identity_cache, a gem that provides query caching with automatic cache invalidation; however, it is affected by most of the other issues we want to address when caching queries at the application-level. You can learn more about the challenges with using the Rails low-level caching API or other caching technologies such as IdentityCache here.

IdentityCache is deliberately opt-in for all call sites that want to use caching. In comparison, RedisMemo is still deliberate in that clients should specify what computation and models should be cached. However, when caching does make sense, RedisMemo makes caching easy and robust by automatically using the cached code paths.