A cache that uses both memory and memcached

Summary

• In Ruby on Rails, minimizing an amount of queries to DB as much as possible is important to scale efficiently
• A caching mechanism can be implemented by using both memcached and memory cache with good cache characteristics of both.

Site data features and caching strategies

• Sites have different caching strategies for different characteristics, such as frequency of data updates
• This site
  Price Sale 最安値でラクラクお買い物

  日本最高品質、完全無料の最安値検索サービスです。メルカリ、楽天、ヤフオクなど50以上サイトから最安値ですぐに購入できます。安心して購入できる未使用品や中古品を厳選して紹介しています。

  Price Sale
  is a media site where it doesn't matter if the information displayed in the view is several hours old.

The most important thing in making Rails faster

• Avoid accessing DB as much as possible.
• When it accesses DB, only required columns are retrieved.
• Reduce the creation of Ruby objects as much as possible.
• It is necessary to understand the above basic structure of the cash flow.

Features of memcached and Memory Cache

Both middleware, such as memcached (Redis), and memory-based caching mechanisms have their own characteristics.

memcached

• Ability to share caches between processes using the same memcached.
• If you have a lot of access to memcached, it's a bottleneck sometimes.
• Caching complex objects in memcached takes a long time to serialize and deserialize.
• It is preferable to specify this in Rails.cache because it can be shared by multiple processes.
• For an overview of the implementation, I recommend
  Caching with Rails: An Overview — Ruby on Rails Guides

  Caching with Rails: An OverviewThis guide is an introduction to speeding up your Rails application with caching.Caching means to store content generated during the request-response cycle and to reuse it when responding to similar requests.Caching is often the most effective way to boost an applicati…

  Ruby on Rails Guides
  as it is written in an easy to understand manner.

Memory cache (memoist)

• Per-process caches; memory caches are not shared with other processes
• Fast read/write with little to no bottleneck
• GitHub - matthewrudymemoist: ActiveSupport::Memoizable with a few enhancements

  ActiveSupport::Memoizable with a few enhancements. Contribute to matthewrudymemoist development by creating an account on GitHub.

  GitHub
  is one of the good libraries for memory cache.

Cache that uses both memory and memcached

The cache structure is as follows:

• Using memcached and memory cache together to improve performance
• Periodically store the results of asynchronous calculations in memcached. If there is no calculation result, the value is retrieved from DB.
• When memcached is accessed, it is stored in the memory cache and the contents of the memory cache are read the second time around.
• Refresh the memory cache every hour and the memcached every hour.

This allows you to share the cache while limiting the amount of access to the memcached.

In addition, frequently accessed data can take advantage of the fast response of the memory cache.

Increase the cache hit rate for memcadched

• The most important point in this strategy is to properly refresh the memcached, memory cache.
• To increase the cache hit rate of memcached, it asynchronously computes and refreshes the calculation results.

Make the cache a regular refresh mechanism

memcached

• Periodically perform batch processing with the rake task and store the calculation results in memcached.
• Refresh the cache periodically at that time.

Memory Cache(memoist)

The process of web application (PUMA) implements a mechanism to clear the cache after a certain time elapses in after_action of application_controller.rb, and it clears the cache periodically.

In the process of an active job, a mechanism to clear the cache is implemented in after_perform of application_job.rb after a certain time elapses, and the cache is cleared periodically.

Notes on caching

• The cache key should be unique. If you use an array as a cache key, be careful not to create a sort order or empty elements.