README

WordPress plugin to store user roles and user levels in a taxonomy, for performance.

Having many users in a WordPress database leads to some bad performance when it comes to quering for users by role, user level, or counting users. This is particularly a problem because the default WordPress admin users list table, post table and post edit screens all make these queries.

The fundamental issue with the default WordPress storage mechanism is the use of Post Meta for user levels and roles. To add insult to injury, roles are stored as a serialized array, so any queries for users by role results in a MySQL LIKE query on an unindexed text column (meta_value).

Roles to Taxonomy registers two shadow taxonomies to associate user objects with role/user level terms. This results in a much faster lookup for users in a given role. The term count field is also be used to calculate user counts in some cases.

Performance Comparisons (2.2 million network users, 1.5 million on a single site)

Usage

When you activate the plugin, you will need to backfill the taxonomies with the existing user roles and levels. The plugin comes with a WP CLI command to perform a synchronization of existing roles.

wp roles-to-taxonomy sync [--verbose] [--batch-size=<number>] [--progress] [--offset=<number>] [--fast-populate] [--limit=<number>]

• batch-size=x dictates the chunks of users to process at a time. If you have 10,000+ users, you'll want to set this to reasonable chunks (say 5,000) to not exhaust all available memory.
• progress will output a progress bar.
• limit=x will restrict the total amount of user roles synced to taxonomies.
• offset=x will resume updating users from a given offset number.
• fast-populate will enable a very fast term/taxonomy population mode. This uses many direct SQL queries so bypass WordPress process, hooks, object cache etc. This is only recommended if you do not already have any users synced to taxonomies. This should only be used when you have 100,000+ users. In testing, it is able to insert use role and level terms at a rate of 1 million users per 15 minutes.