High Performance Drupal

Chapter 16. Alternative Storage and Cache Backends

Drupal uses MySQL as the default storage backend for all data on a site. This includes low-level APIs such as caching and queues, as well as site content.

MySQL provides a great database backend for Drupal, but there are certain issues that arise, especially as Drupal sites get larger. Drupal is very modular and flexible in terms of how data can be defined. Entity types such as nodes, users, and comments can be extended with configurable fields, and then the Views module allows those to be queried and presented—all without leaving the user interface. This is a very powerful set of features, but it can make getting great performance out of a traditional relational database quite difficult. The entity/field storage provided by core provides one or two tables for each core entity, then two tables for each individual field (one for the current or “live” revision and one for older revisions). Queries with a condition on one field that sort on another then have to join across multiple tables, which is very hard to optimize with a relational database.

Another common performance bottleneck that particularly affects higher-traffic Drupal sites is that cache and session data are stored in MySQL tables by default. While this is generally fine for smaller sites, as the number of cache and session objects grows, frequent actions such as cache invalidation put a high demand on the database layer. This can cause increased network traffic and ultimately ...