Hi All,
First of all thank you for such great product like InfluxData.
I know that question batch against stream was raised several times, but will try to ask it again in regards to the following task.
In some cases (actually it’s pretty often) you need to worm-up pipeline - feed it with batch of data. For example you need to calculate standard deviation at from now()-N to now(). Or you need warm-up for example high pass filter (or low pass filter) - in any cases you need to query some historical data (batch mode), feed it through pipeline, warm-up some internal states, ignore output of that warm-up data and switch to stream mode to process data normally.
It can be done either in pure stream mode, but you need to collect points for certain time, and some time it’s not acceptable when pipeline in warm-up mode for 30 minutes, or it can be done in pure batch mode - but it will not be efficient because you need to recalculate all metrics characteristics with every batch.
What would be my option to address such problem, is there solution that allow to approach problem in flexible way ?
© 2023 InfluxData, Inc.
