Azure Messaging, Queueing, and Events - Which service is right for your solution? - Brian Gorman

Azure provides several messaging and event services, each optimized for different scenarios:

• Event Hub - For big data streaming, IoT telemetry, millions of events per second
• Event Grid - For reactive programming, event monitoring and responding to state changes
• Service Bus - For enterprise messaging, ordered delivery, transactions
• Storage Queues - For simple queuing needs with large storage requirements (>80GB)

Key distinctions between services:

• Messages vs Events - Messages contain raw data and state that needs processing, Events are notifications something happened
• At-most-once vs At-least-once delivery
• Ordering guarantees (Service Bus can guarantee order with sessions)
• Storage capacity (Storage Queues >80GB, Service Bus <80GB)
• Throughput requirements (Event Hub handles millions/sec)

Event Hub specifics:

• Uses partitions for scale and ordered processing
• Consumer groups enable parallel processing
• Capture feature for cold storage path
• Throughput Units control ingress/egress capacity

Service Bus features:

• Topics and subscriptions for pub/sub
• Message sessions for ordered processing
• Dead letter queues
• Message TTL and scheduling
• RBAC and SAS token authentication

Best practices:

• Use AMQP instead of HTTPS for better performance
• Consider message size limits (64KB for Storage Queues)
• Enable auto-capture for Event Hubs when needed
• Use managed identities over connection strings
• Implement proper error handling and dead letter queues

Integration capabilities:

• Logic Apps for event-driven workflows
• Stream Analytics for real-time analytics
• Functions for serverless processing
• IoT Hub compatibility with Event Hub SDK