Azure IoT Hub单设备消息接收限制、无额外设备扩容方案及吞吐量单元调整未生效问题咨询

Let's break down your questions and observations step by step:

1. Does Azure IoT Hub have per-device message limits? #

Yes, Azure IoT Hub enforces per-device message throttling limits to prevent a single device from consuming excessive hub resources. For the MQTT protocol (which you're using), the default limits for standard-tier IoT Hubs are:

• 100 messages per second (msg/sec) per device
• 6000 messages per minute (msg/min) per device

Your initial test result of ~3.6k msg/min aligns with hitting a per-device constraint (likely due to minor client-side throttling or environmental overhead in your test setup). When you added a second device, you distributed the load across two logical identities, bypassing the single-device cap and unlocking more of your hub's total throughput.

2. Can I remove or increase this limit without registering additional devices? #

Unfortunately, no—this per-device throttling is a built-in safeguard in IoT Hub to ensure fair resource allocation across all connected devices. There’s no way to bypass or raise this limit for a single device identity directly.

If your production environment can only use one physical device, you have a couple of practical workarounds:

• Register multiple device identities for the same physical device: Create multiple IoT Hub device identities and rotate them in your client code to split message traffic across logical identities. This lets a single physical device act as multiple logical devices to avoid hitting the per-identity limit.
• Leverage Azure IoT Edge: If your device runs IoT Edge, you can use the Edge hub to batch or aggregate messages before sending to the cloud. While this doesn’t bypass the per-device identity limit, it optimizes message delivery and can help reduce the number of individual messages sent per second.

3. Why didn't increasing throughput units (TUs) affect my initial test results? #

Throughput units (TUs) define the overall maximum throughput capacity of your entire IoT Hub, not per-device limits. For standard-tier hubs:

• 1 TU supports up to 1000 msg/sec ingress (device-to-cloud) and 1000 msg/sec egress (cloud-to-device)
• 20 TUs would theoretically support 20,000 msg/sec ingress

In your initial single-device tests, you hit the per-device throttling limit first (~100 msg/sec = 6000 msg/min, or slightly lower in your environment), not the hub’s TU-based total capacity. Increasing TUs doesn’t modify per-device constraints, so you saw no improvement. Once you added a second device, you could utilize more of the hub’s TU capacity because you were no longer bottlenecked by the single-device cap.

Additional Tips for Load Testing #

• To reach your target of 18,000 msg/sec, you’ll need at least 180 device identities (since each can handle ~100 msg/sec) paired with a minimum of 18 TUs (18*1000 = 18,000 msg/sec)
• Ensure your client code handles connection pooling and retries properly to avoid client-side throttling
• Monitor IoT Hub metrics like d2c.messages.throttled in the Azure Portal to confirm when you’re hitting limits

内容的提问来源于stack exchange，提问作者Shunsuke Hioki