Per-aggregate ordering is what we really want. But queues with competing consumers make it surprisingly easy to break. This article shows how the “fix” naturally evolves into a saga.

MilanJovanovic's platform  covers topics related to software development, technology trends, and coding tutorials. Through articles, tutorials, and code examples, MilanJovanovic offers insights into programming languages, development frameworks, and best practices in software engineering. Readers can learn about software design principles, development methodologies, and emerging technologies to enhance their coding skills and build high-quality software applications.

Milan Jovanović

Per-aggregate message ordering is essential in distributed systems, but competing consumers break this guarantee. The article walks through the evolution from domain events to the Outbox pattern for reliability, then addresses ordering challenges. It demonstrates how enforcing sequential processing per aggregate naturally leads to choreographed sagas, and eventually to orchestrated state machine sagas when control and observability become critical. Broker-level ordering features (like Kafka partitions or SQS FIFO) help but don't eliminate the need for workflow patterns like Outbox, Inbox, and compensation logic.

Solving Message Ordering from First Principles

Domain Events Feel Like the Clean Solution

The Outbox Makes Publishing Reliable (but not ordered)

Competing consumers Are Great, Until Order Matters

What We Really Want is Per-Aggregate Ordering

A Single Consumer Solves Ordering But Limits Scale

Publish the Next Message From the Handler

Congratulations, You Built a Choreographed Saga

If You Want Control, Introduce a State Machine Saga

Broker Support Helps with Ordering, not Correctness