TechPrufer
Get started
Shopify
Shopify logoShopify · Feature trail

ActiveRecord Model Caching

2 posts · Apr 2013 – Sep 2021

What this trail covers

Shopify has evolved its caching mechanisms to improve performance and reduce database load. This includes developing custom write-through caches for high-traffic features like the Shop app's home feed, utilizing Memcached for its simplicity and low overhead. The system handles concurrent updates and ensures cache validity through a pending writes key and short-expiry keys, making changes transparent to developers via Active Record Concerns. Rollout strategies involve phased testing and comparison of cached vs. database data to guarantee correctness. This has led to significant reductions in database load, app latency, and overall CPU usage.

All Shopify trails
  1. Making Shopify’s Flagship App 20% Faster in 6 Weeks Using a Novel Caching Solution - ShopifyThis post details the development of a custom write-through cache for the Shop app's home feed using Memcached. The solution addresses challenges with concurrent updates by implementing a pending writes key and a short-expiry key to manage cache validity. Active Record Concerns were used to integrate the caching logic transparently into the existing ORM. A phased rollout strategy was employed, including a period where cached and database data were compared to ensure correctness before global deployment.
  2. IdentityCache: Improving Performance one Cached Model at a Time - ShopifyThis post introduces IdentityCache, an open-sourced library for caching ActiveRecord models using Memcached. It implements a read-through caching strategy with explicit expiration via after_commit hooks, reducing database load. The library supports caching single objects, associations (has_many, has_one, belongs_to), and embedding associations within parent objects to further reduce Memcached calls. It also allows for custom cache indexes based on model fields.