I have some highly variable time series data and are looking for an efficient way to store it. It only needs to be queried based on timestamp and two tags (type and device).
A device typically logs a set of 15-20 values (from a possible pool of around 200) and this set of values will generally never change over the entire life of the device, so it seems inefficient to have all these extra fields(columns) for values I'll likely never use.
Because of the above, I'm considering protocol buffers as an efficient way to store this data, but struggling with a way to store this data in influxDB.
Is it possible to store either protocol buffers OR binary data in InfluxDB? I really don't want to use an encoding scheme (eg: base64) to store it as a string.