Azure Event Hubs · Rate Limits

Microsoft Azure Event Hubs Rate Limits

Event Hubs throughput is bounded by the purchased Throughput Units (Basic/Standard) or Processing Units (Premium). Each TU allows up to 1 MB/s or 1,000 events/s ingress, 2 MB/s egress. Quota exceeded triggers ServerBusy. Premium PUs and Dedicated CUs scale higher; Auto-Inflate on Standard scales TUs up automatically.

Microsoft Azure Event Hubs Rate Limits is the machine-readable rate-limit profile for Azure Event Hubs on the APIs.io network, conforming to the API Commons Rate Limits specification.

It captures 7 rate-limit definitions, measuring bandwidth, throughput_units, processing_units, event_size, and connections.

The profile also includes 4 backoff/retry policies defined and response codes documented for serverBusy and throttled.

Tagged areas include Rate Limiting, Event Streaming, and Microsoft Azure.

7 Limits Throttle: 429
Rate LimitingEvent StreamingMicrosoft Azure

Limits

Per-TU ingress (Basic, Standard) namespace
bandwidth · second
1 MB/s or 1,000 events/s per TU
Per-TU egress (Basic, Standard) namespace
bandwidth · second
2 MB/s per TU
Standard tier max TUs (self-serve) namespace
throughput_units
20
Standard namespaces can request up to 40 TUs via support.
Standard tier max TUs (with support request) namespace
throughput_units
40
Premium PUs per namespace namespace
processing_units
16
Purchase in 1, 2, 4, 6, 8, 10, 12, or 16 PUs. Each PU provides higher and more isolated throughput than a TU.
Ingress event size per_request
event_size
1 MB (Standard); larger on Premium/Dedicated
Brokered connections namespace
connections
Basic 100; Standard 1,000; Premium/Dedicated higher

Policies

Auto-Inflate
Standard namespaces can enable Auto-Inflate to scale TUs upward automatically when ingress hits the per-TU ceiling, up to a configured maximum.
Backoff on ServerBusy
When the broker returns ServerBusy (50008) or 429, back off exponentially with jitter and reduce send concurrency.
Partition planning
Use partitions to scale consumer parallelism; default 32 partitions per Event Hub on Standard, more on Premium/Dedicated.
Batch sends
Batch events to amortize overhead and stay within per-TU/PU event/sec ceilings.

Sources