Dear community I have recently been task to build a cloud architecture for a vehicle telemetry system, preferably in AWS. There are going to be around 50-100 vehicles sending JSON messages via MQTT to a IoT Core broker every 10 seconds, each message weighting 200 bytes max (5gB of new data each month aprox). Now I need to setup a frontend capable of showing realtime/near-realtime data of the sensor as well as a summary of aggregate data (average fuel consumption, timeseries graphs, GPS locations) of the last day, week, month and semester. I have done a bunch of research and found about telemetry architecture and it usually have a Streaming branch and a Batch one, but then I lack the experience to decide how elaborated should my setup be, as maybe for the amount of data I would handle may be supported by a simpler architecture, also I'm aware that historical data queries can get very expensive overtime and they will be requested a lot by the clients.
Here are some architectures ideas I have come up. Timestream seems tempting but it seems queries are expensive and I cannot delete data. I have also seen the use of Athena with S3 buckets.
My question then is which architecture features it is recommended I use given my data flow and requirements.
We don't spare a lot of resources so a cheaper setup is preferable, it is OK to average out a batch of records in the historical stream (let's say average the last 6 records to be a minute) and ideally to be scalable to more vehicles.
Lastly I would love to listen your route of thoughts at the moment to validate your architectures. Are there any open source technologies I should check out? As always thank you very much for the help!
