Azure Event Hubs

Use it when

• Your platform is Azure and you want managed streaming that Stream Analytics, Functions, and Data Explorer plug into natively.
• Existing Kafka producers and consumers should connect unchanged, without a Kafka cluster existing anywhere; the hub speaks the Kafka protocol from the Standard tier up.
• High-volume telemetry and log ingestion is the job, with Capture continuously landing the stream in Blob Storage or Data Lake without a connector.
• Ops capacity is thin; tiers and throughput units replace brokers and rebalancing.

Think twice when

• You need long replay windows on a budget: retention is 7 days on Standard and 90 days on Premium and Dedicated.
• Cross-region data replication matters; Standard offers metadata-only disaster recovery, and data replication starts at Premium.
• The workload is transactional enterprise messaging or reactive routing; Microsoft points those at Service Bus and Event Grid instead.

How it runs

Fully managed: pick a tier, get event hubs (Kafka topics by another name) divided into partitions with consumer groups tracking offsets. Auto-inflate scales throughput units with demand; Dedicated gives reserved clusters. Billing is throughput or processing units plus ingress. Transformations happen downstream in Stream Analytics or Functions.