This document discusses Gannett's transition from synchronous to asynchronous indexing using queues. It summarizes Gannett's current indexing architecture, the problems with synchronous operations like outages during schema changes, and how moving to a queue-based system with RabbitMQ allows for faster schema changes, auto-scaling of workers, and more resilient indexing with minimal authoring impact. The queue-based approach includes real-time, batch, and prep queues to handle different types of document changes and updates in a decoupled, eventually consistent manner.