Azure Messaging

David Watson
Cloud Solution Architect

Title

Agenda

Choice
Messages vs Events
Storage Queues
Azure Storage Queues
Service Bus
Azure Service Bus
Event Hubs
Azure Event Hubs
Event Grid
Azure Event Grid
Guide
Choosing the Right Service

Messages vs Events

Messages

  • Raw data produced by one service to be consumed by another
  • Contains the data itself, not just a reference
  • Sender has an expectation about how the message is handled
  • A contract exists between sender and receiver

Events

  • Lightweight notification that something happened
  • Contains metadata, not necessarily the full data payload
  • Publisher has no expectation about handling
  • Zero or many subscribers can react independently

Core Messaging Patterns

Point-to-Point (Queue)

  • One sender, one receiver
  • Competing consumers pattern
  • Message consumed once
  • Load leveling & decoupling

Publish / Subscribe

  • One publisher, many subscribers
  • Each subscriber gets a copy
  • Fan-out distribution
  • Subscriber filtering

Azure Messaging Services

Storage Queues
Storage Queues
Simple, high-volume queuing
Service Bus
Service Bus
Enterprise messaging broker
Event Hubs
Event Hubs
Big data streaming
Event Grid
Event Grid
Reactive event routing

Azure Storage Queues

Simple, scalable message queuing

Azure Storage Queues

What Is It?

  • Simple queue built on Azure Storage
  • Part of every storage account
  • REST-based access anywhere
  • Messages up to 64 KB

Key Characteristics

  • Queue can exceed 80 GB total
  • Server-side logs of all transactions
  • At-least-once delivery
  • TTL up to 7 days (or unlimited)

When to Use Storage Queues

  • Simple work queue / job dispatch
  • Queue size may exceed 80 GB
  • You need server-side transaction logs
  • Minimal feature requirements (no ordering, sessions, dead-letter)
  • Cost-sensitive workloads with high message volume

Azure Service Bus

Enterprise-grade message broker

Azure Service Bus

  • Fully managed enterprise message broker
  • Supports queues (point-to-point) and topics (pub/sub)
  • Messages up to 256 KB (Standard) / 100 MB (Premium)
  • AMQP 1.0, HTTP/REST, and client SDK access
  • Tiers: Basic → Standard → Premium

Service Bus Queues

How It Works

  • FIFO delivery with sessions
  • Competing consumers pattern
  • Peek-lock or receive-and-delete
  • Dead-letter queue (DLQ)

Delivery Guarantees

  • At-least-once (peek-lock)
  • At-most-once (receive-and-delete)
  • Duplicate detection
  • Message deferral

Service Bus Topics & Subscriptions

Publish / Subscribe

  • Publisher sends to a topic
  • Each subscription gets a copy
  • Up to 2,000 subscriptions per topic
  • Works with all queue features

Subscription Filters

  • SQL filter — match on properties
  • Correlation filter — match specific fields
  • Boolean filter — true/false (all or none)
  • Filter actions can modify message properties

Service Bus Advanced Features

⏱️
Scheduled Delivery
Delay message processing
🔄
Auto-Forwarding
Chain queues & topics
💀
Dead-Lettering
Isolate failed messages
🔁
Duplicate Detection
Idempotent processing
📦
Sessions
Ordered & grouped delivery
🌍
Geo-DR
Cross-region failover

Service Bus Tiers

Feature Basic Standard Premium
Queues ✓ ✓ ✓
Topics / Subscriptions — ✓ ✓
Max Message Size 256 KB 256 KB 100 MB
Dedicated Resources — — ✓
VNet / Private Endpoint — — ✓
Geo-Disaster Recovery — — ✓

Azure Event Hubs

Big data streaming platform

Azure Event Hubs

  • Fully managed event streaming platform
  • Millions of events per second at low latency
  • Apache Kafka compatible (Kafka endpoint)
  • Retention: 1–90 days (Standard) / up to unlimited (Premium/Dedicated)
  • Designed for telemetry, logging, and analytics pipelines

Event Hubs Key Concepts

📊
Partitions
Parallel event streams for scale
👥
Consumer Groups
Independent views of the stream
📸
Capture
Auto-archive to Blob / Data Lake
⚡
Throughput Units
1 TU = 1 MB/s in, 2 MB/s out
🕐
Event Retention
Time-based message store
📍
Checkpointing
Track consumer position

Event Hubs & Apache Kafka

Kafka Compatibility

  • Kafka producer/consumer APIs work directly
  • No Kafka cluster to manage
  • Standard tier and above
  • Kafka Connect & Streams supported

When to Use Kafka Endpoint

  • Migrating existing Kafka workloads
  • Multi-cloud / hybrid scenarios
  • Leveraging Kafka ecosystem tools
  • Avoiding operational overhead

Azure Event Grid

Reactive, event-driven architectures

Azure Event Grid

  • Fully managed event routing service
  • First-class support for Azure service events
  • React to state changes with near real-time delivery
  • Push-based delivery model (HTTP webhooks, Functions, etc.)
  • Pay per event — no idle cost

Event Grid Concepts

Event Sources

  • Azure Blob Storage
  • Resource Groups / Subscriptions
  • IoT Hub, Service Bus, Event Hubs
  • Custom topics (your apps)

Event Handlers

  • Azure Functions
  • Logic Apps
  • Webhooks (any HTTP endpoint)
  • Storage Queues, Service Bus, Event Hubs

Event Grid Features

🔍
Event Filtering
Subject, type, or advanced filters
🔄
Retry & Dead-Letter
Built-in retry policies
📡
Fan-Out
Multiple subscribers per topic
🔐
Security
Webhook validation & Azure AD
📋
CloudEvents
CNCF standard event schema
⏱️
High Availability
Built-in zone redundancy

Service Comparison

Storage Queues Service Bus Event Hubs Event Grid
Pattern Queue Queue + Pub/Sub Streaming Reactive events
Delivery At-least-once At-least / At-most At-least-once At-least-once
Protocol HTTP/REST AMQP, HTTP AMQP, Kafka, HTTP HTTP (push)
Throughput High Medium–High Very High Very High
Ordering No FIFO (sessions) Per partition No
Max Msg Size 64 KB 256 KB – 100 MB 1 MB (256 KB std) 1 MB

Choosing the Right Service

Storage Queues
Storage Queues
Simple decoupling, large backlogs, cost-effective
Service Bus
Service Bus
Enterprise messaging, ordering, transactions, DLQ
Event Hubs
Event Hubs
High-throughput streaming, telemetry, Kafka migration
Event Grid
Event Grid
Event-driven reactions, Azure service integration

Decision Framework

  • Need ordered delivery or transactions? → Service Bus
  • Millions of events/second at low latency? → Event Hubs
  • React to Azure resource changes? → Event Grid
  • Simple queue with no frills? → Storage Queues
  • Combine services — e.g., Event Grid triggers a Function that writes to Service Bus

Key Takeaways

  • Messages vs Events — understand the distinction to choose correctly
  • Storage Queues — simple, cost-effective, massive scale
  • Service Bus — enterprise broker with rich feature set
  • Event Hubs — big data streaming, Kafka-compatible
  • Event Grid — reactive event routing, push-based

Questions?

Thank you!