Caching in Ruby on Rails serves as a vital performance optimization technique that can dramatically improve application response times. I’ve spent years implementing various caching strategies, and I’ll share the most effective techniques I’ve discovered.

Multi-level Cache Architecture

A multi-level cache architecture combines different caching layers to optimize data retrieval. The most common implementation uses memory-based caching (like Memcached) as the first layer and disk-based caching as the backup.

class MultiLevelCache
  def fetch(key)
    MemoryStore.fetch(key) do
      DiskStore.fetch(key) do
        yield
      end
    end
  end
end

Cache key design remains crucial for effective caching. Complex keys should incorporate relevant model attributes and timestamps.

def cache_key_generator(record)
  "#{record.class.name}/#{record.id}-#{record.updated_at.to_i}"
end

Cache warming prevents the initial slowdown when populating empty caches. Background jobs can pre-populate frequently accessed data.

class CacheWarmer
  include Sidekiq::Worker

  def perform(resource_type)
    resources = resource_type.constantize.frequently_accessed
    resources.each do |resource|
      Rails.cache.write(resource.cache_key, resource)
    end
  end
end

Version-based invalidation offers a clean approach to managing cache updates:

class VersionedCache
  def fetch_with_version(key)
    version = current_version(key)
    versioned_key = "#{key}/v#{version}"
    
    Rails.cache.fetch(versioned_key) do
      yield
    end
  end

  private

  def current_version(key)
    Rails.cache.fetch("#{key}/version") { 1 }
  end

  def increment_version(key)
    Rails.cache.increment("#{key}/version")
  end
end

Cache store selection impacts performance significantly. Redis offers advanced features like automatic key expiration and atomic operations:

config.cache_store = :redis_cache_store, {
  url: ENV['REDIS_URL'],
  expires_in: 1.day,
  race_condition_ttl: 10,
  error_handler: -> (error) { Rails.logger.error(error) }
}

Race condition prevention becomes essential in high-traffic applications:

class SafeCache
  def fetch_with_lock(key)
    Rails.cache.fetch(key, race_condition_ttl: 10.seconds) do
      with_lock(key) do
        yield
      end
    end
  end

  private

  def with_lock(key)
    lock_key = "#{key}_lock"
    acquired = Rails.cache.write(lock_key, true, unless_exist: true, expires_in: 30.seconds)
    
    if acquired
      begin
        yield
      ensure
        Rails.cache.delete(lock_key)
      end
    else
      sleep 0.1 until Rails.cache.read(lock_key).nil?
      Rails.cache.read(key)
    end
  end
end

Cache monitoring helps identify performance bottlenecks:

class CacheMonitor
  def track_cache_hit(key)
    StatsD.increment('cache.hit', tags: ["key:#{key}"])
  end

  def track_cache_miss(key)
    StatsD.increment('cache.miss', tags: ["key:#{key}"])
  end

  def track_cache_write(key)
    StatsD.increment('cache.write', tags: ["key:#{key}"])
  end
end

Fragment caching optimizes partial view rendering:

class ProductsController < ApplicationController
  def index
    @products = Product.all
  end
end

# View template
<% cache_if user_signed_in?, @products do %>
  <% @products.each do |product| %>
    <% cache [product, current_user&.admin?] do %>
      <%= render partial: 'product', locals: { product: product } %>
    <% end %>
  <% end %>
<% end %>

Russian Doll caching nests cache fragments:

class Comment < ApplicationRecord
  belongs_to :post, touch: true
end

class Post < ApplicationRecord
  has_many :comments
end

# View template
<% cache post do %>
  <%= render post %>
  <% post.comments.each do |comment| %>
    <% cache comment do %>
      <%= render comment %>
    <% end %>
  <% end %>
<% end %>

Query caching reduces database load:

class QueryCache
  def fetch_records
    Rails.cache.fetch('expensive_query', expires_in: 1.hour) do
      Product.joins(:category)
             .where(active: true)
             .includes(:variants)
             .to_a
    end
  end
end

HTTP caching implementation:

class ProductsController < ApplicationController
  def show
    @product = Product.find(params[:id])
    fresh_when(@product)
  end
end

# Alternative with ETag
class ArticlesController < ApplicationController
  def show
    @article = Article.find(params[:id])
    if stale?(etag: @article, last_modified: @article.updated_at)
      render @article
    end
  end
end

Counter caching improves count queries:

class Post < ApplicationRecord
  belongs_to :author
  belongs_to :category, counter_cache: true
end

class AddCounterCacheToCategories < ActiveRecord::Migration[6.1]
  def change
    add_column :categories, :posts_count, :integer, default: 0
    
    reversible do |dir|
      dir.up { data }
    end
  end

  def data
    Category.find_each do |category|
      Category.reset_counters(category.id, :posts)
    end
  end
end

Implementing cache sweepers:

class ProductSweeper < ActionController::Caching::Sweeper
  observe Product

  def after_save(product)
    expire_cache_for(product)
  end

  def after_destroy(product)
    expire_cache_for(product)
  end

  private

  def expire_cache_for(product)
    expire_page(controller: 'products', action: 'show', id: product.id)
    expire_fragment("product_#{product.id}")
  end
end

Low-level cache implementation:

class LowLevelCache
  def read_multi(*keys)
    Rails.cache.read_multi(*keys)
  end

  def write_multi(hash)
    hash.each do |key, value|
      Rails.cache.write(key, value)
    end
  end

  def fetch_multi(*keys)
    results = read_multi(*keys)
    missing = keys - results.keys

    if missing.any?
      new_values = yield(missing)
      write_multi(new_values)
      results.merge!(new_values)
    end

    results
  end
end

These caching techniques significantly improve Rails application performance when implemented correctly. Regular monitoring and adjustment ensure optimal cache effectiveness as application needs evolve.