Question

What do you think is a data store appropriate for sensor data, such as temperature, humidity, velocity, etc. Users will have different sets of fields and they need to be able to run basic queries against data.

Relational databases like MySQL is not flexible in terms of schema, so I was looking into NoSql approaches, but not sure if there’s any particular project that’s more suitable for sensor data. Most of NoSql seem to be geared toward log output as such.

Any feedback is appreciated.

Thanks!

Answer 1

I still think I would use an RDBMS simply because of the flexibility you have to query the data. I am currently developing applications that log approximately one thousand remote sensors to SQL Server, and though I’ve had some growing pains due to the “inflexible” schema, I’ve been able to expand it in ways that provides for multiple customers. I’m providing enough columns of data that, collectively, can handle a vast assortment of sensor values, and each user just queries against the fields that interest them (or that their sensor has features for).

That said, I originally began this project by writing to a comma separated values (CSV) file, and writing an app that would parse it for graphing, etc. Each sensor stored its data in a separate CSV, but by opening multiple files I was able to perform comparisons between two or more sets of sensor data. Also CSV is highly compressible, opens in major office applications, and is fairly user-editable (such as trimming sections that you don’t need).