Design of a Distributed Messaging Queue: Part 1

The frontend must validate a request, identify its queue, and route it to the current owner. Keep metadata and deduplication decisions separate from the broker’s durable message acceptance so a retry has a well-defined outcome.

We have discussed the requirements and design considerations. Now, let’s examine the high-level design of a distributed messaging queue.

Distributed messaging queue

A distributed queue adds replicas and ownership management around message storage. These mechanisms can cover failures and add capacity, but their acknowledgement and failover rules must be specified before claiming durability or availability.

High-level design

To simplify the design discussion, we assume the following: