Solid Cache Coder

Overview

Solid Cache is Rails 8's default cache backend—a database-backed cache store using NVMe storage instead of RAM. It eliminates the need for Redis/Memcached while providing FIFO eviction.

Configuration

Basic Setup

config/environments/production.rb

config.cache_store = :solid_cache_store

Advanced Configuration

config.cache_store = :solid_cache_store, { database: :cache, expires_in: 2.weeks, size_estimate: 500.megabytes, namespace: Rails.env, compressor: ZSTDCompressor }

Database Setup

Separate Cache Database (Recommended)

config/database.yml

production: primary: <<: *default url: <%= ENV["DATABASE_URL"] %> cache: <<: *default url: <%= ENV["CACHE_DATABASE_URL"] %> migrations_paths: db/cache_migrate

Run Migrations

bin/rails solid_cache:install:migrations bin/rails db:migrate

Cache Key Design

Hierarchical Keys

Rails.cache.fetch("user:#{user.id}:profile:#{profile.id}") { expensive_computation } Rails.cache.delete_matched("user:#{user.id}:*") # Pattern-based deletion

Russian Doll Caching

<% cache @post do %> <%= @post.body %> <% @post.comments.each do |comment| %> <% cache comment do %><%= render comment %><% end %> <% end %> <% end %>

Expiration Strategies

Per-entry expiration

Rails.cache.write("session:#{id}", data, expires_in: 30.minutes)

Fetch with expiration

Rails.cache.fetch("expensive_query", expires_in: 15.minutes) { ExpensiveQuery.run }

Performance Tuning

Enable ZSTD Compression

config.cache_store = :solid_cache_store, { compressor: ZSTDCompressor }

Batch Operations

keys = users.map { |u| "user:#{u.id}:preferences" } results = Rails.cache.read_multi(*keys)

Rails.cache.fetch_multi(*keys) { |key| compute_value_for(key) }

Database Maintenance

lib/tasks/cache_maintenance.rake

namespace :cache do task vacuum: :environment do ActiveRecord::Base.connected_to(database: :cache) do ActiveRecord::Base.connection.execute("VACUUM ANALYZE solid_cache_entries") end end end

Solid Cache vs Redis

Choose Solid Cache When Choose Redis When

Simplifying infrastructure Sub-millisecond latency critical

Using PostgreSQL/SQLite as primary Extremely large cache working set

Single-server or small clusters Need pub/sub or complex invalidation

Preferring NVMe over RAM caching Already running Redis

Anti-Patterns

Anti-Pattern Problem Solution

Single database at scale Resource contention Separate cache database

No compression Wasted storage Enable ZSTD

Infinite expiration Unbounded growth Set reasonable max_age

No maintenance Table bloat Schedule VACUUM

Output Format

When configuring Solid Cache, provide:

• Database Setup - Multi-database configuration

• Cache Config - Store options and expiration

• Performance - Compression and pool settings

• Maintenance - Cleanup tasks and schedules