Azure Event Hubs is Microsoft's managed event-ingestion and streaming service. It takes in large volumes of events - telemetry, logs, clickstreams - and makes them available to downstream consumers, with a fully managed, partitioned, log-based design. Notably, it exposes a Kafka-compatible endpoint, so Kafka clients can produce and consume from Event Hubs with only a connection-string change.

The model

An Event Hub is a partitioned log, conceptually like a Kafka topic: events are appended to partitions and retained for a configured period, and consumers read at their own offset. Consumers organize into consumer groups, each with an independent view of the stream. Throughput on the standard tiers is measured in throughput units (or processing units on Premium/Dedicated) that you scale to match ingress and egress volume.

Kafka compatibility

The Kafka endpoint means existing Kafka producers, consumers, and many connectors work against Event Hubs without code changes - you swap the bootstrap servers and credentials. This lets teams adopt a managed Azure service while keeping portable, open Kafka client code. (Some Kafka ecosystem features like full Kafka Connect/Streams parity have limits; check current support for advanced cases.)

# A standard Kafka client pointed at Event Hubs
bootstrap.servers = NAMESPACE.servicebus.windows.net:9093
security.protocol  = SASL_SSL
sasl.mechanism     = PLAIN   # connection string as the password

Capture and integration

A standout feature is Event Hubs Capture: automatically write incoming events to ADLS or Blob storage as Avro files on a time/size trigger, with no consumer to run - instant archival of the raw stream to the lake. Event Hubs also feeds Azure Stream Analytics, Functions, and Databricks for processing, and is the ingestion front door for many Azure analytics pipelines.

Where it fits

  • High-volume telemetry/IoT and log ingestion on Azure (it underpins parts of Azure Monitor).
  • Teams wanting managed streaming with portable Kafka client code on Azure.
  • Pipelines that need automatic raw-stream archival to the lake via Capture.

Event Hubs vs Kafka vs Service Bus

Event Hubs is for high-throughput event streaming (many events, read by multiple consumers, partitioned log). Azure Service Bus is the broker for enterprise messaging - queues, topics, ordering, transactions, dead-lettering - the RabbitMQ-style role. Versus self-managed Kafka, Event Hubs trades some ecosystem depth and control for being fully managed and natively integrated with Azure storage, identity, and analytics. On Azure, it's the default streaming ingestion layer; pick Service Bus instead when you need rich per-message broker semantics.