Still pretty new to Mongo but have developed an IoT proof of concept writing to an Atlas cluster from a device in the field.
My questions now are around Scalability and Reliability.
Is writing directly to my cluster from my device the recommended way to interact for an IoT application? Is this scalable to 100 or 1000 or 10000 devices?
If not, what would you recommend? I am familiar with some of the AWS infrastructure and could envision using their IoT services to push data to a router that would batch it to my clusters. I also have a local MongoDB on the device for buffering so could batch send from devices to the hosted Cluster.
Just stumbled across your post while doing a bit of retro research on IoT architecture patterns. Funny how some of the core questions, like direct writes vs. batching, are still just as relevant now. I’ve worked on a few projects since where writing directly from devices to a cluster sounded elegant in theory but turned out to be a bit fragile at scale, mostly due to network reliability and cost of managing connection states. These days, i’d lean toward having some form of gateway/router layer, even a lightweight one, to help buffer and manage flow control, especially when you’re dealing with thousands of devices.