Latency on Homey Bridge too high for decent use of (Zigbee) motion sensors?

I’ve been using the Homey Bridge since its Beta introduction, and was hoping that when it moved to ‘Production’, latency would fall to acceptable levels. The reliability increased a LOT, but latency stayed the same, of even got worse over time. I’m located in Belgium and connect to the Stuttgart DC, and when one of my Zigbee sensors see motion, it can take up to 5+ seconds for the Zigbee lamp(s) to switch on (although sometimes it’s less than 1 second so it IS possible). There can even be a time laps of several seconds between two lamps in the same zone. When you’re waiting in the dark, that’s a long time :-). Do any of you experience the same? Is the latency (as I expect) MUCH less on the Pro?

Are those motion sensors connected to a Hue hub by any chance?

No, they’re connected either via an Innr plug or an Innr lamp (1 intermediate device only). In theory they could connect directly to the Bridge, but they prefer to take another route and I don’t think the Athom Bridge/cloud solution can force a certain connection path. I did manage to connect them directly (had to physically shield off the Innr routers with metal plates: even very close to the Bridge the Aqara sensors still preferred the Innr routers). But even then, I didn’t notice a significant differene in response time. And a few hours later, they’d left the Bridge for an Innr router again. I’m afraid that it’s the cloud solution that gives unpredictable latency, which is why I’m asking if others have the same experience.

It sounds like the server your “Homey” is running on is overloaded.

You can test your network latency here: https://support.homey.app/hc/en-us/articles/7494444235292 but my guess is that this show an acceptable figure, because it will only measure network latency and not how fast the server that serves your Homey will process requests.

Homey Pro, being a local solution, will very likely be much faster.

You’re right, I shoudn’t have used the term ‘latency’ which typically refers to the network only, which is not so bad (around 37ms, measured on different occasions in the past). I meant ‘total response time’, which is - in best cases - just below 1s and often way higher. But if it’s the DC server, I’m not sure that I’ll find a ‘Service Level Description’ to rely on :slight_smile: In that case, the bridge/cloud solution is perfect for temperature/humidity… sensors or flows based on time events (e.g. switch on lights at sunset, …) , but not ideal for motion sensors that require a very quick response.

It depends how the docker platform is configured in aws. If they use docker swarm with enough nodes (and powerful vm’s) or paas. Only then you can tell. My latency to main account hosted on aws frankfurt is 23ms, to my other account in london is 27ms. Everything reacts very quick