KubeMQ
Learn

Messaging Patterns

A learning track for messaging on KubeMQ — the fundamentals, the four patterns (Events, Events Store, Queues, RPC), then composing them into architectures.

KubeMQ is a single message broker that supports four distinct messaging patterns — from fire-and-forget broadcasting to reliable point-to-point delivery to synchronous request-reply. This section is a learning track, not just a reference: it starts with vendor-neutral fundamentals, shows how KubeMQ implements each pattern, helps you pick and combine them, then composes them into real architectures.

One broker, four patterns, every use case.

How to read this track

Work through it in order, or jump straight to the tier you need. Each step links to the next.

New to messaging? Start with the Concepts — six short pages that explain the ideas every pattern is built on.

The Four Patterns

The Four Patterns at a Glance

FeatureEventsEvents StoreQueuesRPC
DirectionPub/SubPub/SubPoint-to-PointRequest-Reply
PersistenceNoYes (disk)Yes (disk)No
Delivery guaranteeAt-most-onceAt-least-onceExactly-onceAt-most-once
ReplayNoYes (6 positions)NoNo
AcknowledgmentNoNoYes (manual)Yes (automatic)
OrderingNoYes (sequence)Yes (FIFO)N/A
ResponseNoNoNoYes
Dead letter queueNoNoYesNo
Delayed deliveryNoNoYesNo
CachingNoNoNoYes (queries)

Which Pattern Should I Use?

Use the interactive decision guide to find the right pattern for your use case.

If you need...Use
Lowest latency, message loss acceptableEvents
Pub/sub with no message lossEvents Store
Replay historical messagesEvents Store
One consumer per message, guaranteed deliveryQueues
Retry, DLQ, delayed deliveryQueues
Synchronous request-replyRPC
Execute action, confirm successRPC Commands
Request data, get resultRPC Queries

Combining Patterns

Patterns can work together in a single application. KubeMQ's channel routing syntax lets you publish to multiple patterns simultaneously. Here are two common shapes.

Order pipeline: a command processes the order, emits an event to the durable log, which feeds a queue for reliable fulfillment.

CQRS shape: commands write through the event store, the read service subscribes to build its projection, and clients query that read side.

Channel routing syntax: use ; to separate channels and : to prefix the pattern type:

events:live-feed;events_store:archive;queues:process

Get Started

Was this page helpful?

On this page