I am new to InfluxDB and I am wondering about the schema design I have to use. I have a couple of sensors producing time-stamped data with about 1 Hz. The fields of these sensors are not always the same, still I assume that each field with data is represented by on column.
What causes issues is one sensor type which delivers in the time-stamped data set a 10 MHz sampled voltage amplitude signal (20.000 points, called “rawData”) which is a time series by itself, but the individual points therein are not time stamped, but I could generate time stamps for each of them using the information above.
My question is how I model this in the best way? Do I put the 20.000 points into columns or do I put them into 20.000 rows? If I do the latter, then how to handle all the other data in the time stamped fields - should I repeat them for every row or is there a more elegant way to do that? If this is the way to go, I assume that sensors without
Visualization of the data structure:
- field1: string
- field2: int
- field3: float
- rawData: ndarray[20.000] <-- this is my time series in the time series
- fieldn+1: …