Where|window vs window|where in pipeline

InfluxDB 1 Kapacitor
1

in kapacitor, if we have a stream of data that we’re applying different tasks to, and each task is doing…

stream
    | from().measurement('some_measurement')
    | where(lambda: 'some criteria')
    | window().period('10s').every('1s')
    | count('timestamp')
...

would there be a negative impact, performance-wise, in changing this to

stream
    | from().measurement('some_measurement')
    | window().period('10s').every('1s')
    | where(lambda: 'some criteria')
    | count('timestamp')
...

Behaviorally this would make it so that when there are no matches to ‘some criteria’ a count of 0 would be emitted so we can take action when there are zero matches (assuming the stream itself is noisy enough to have non-matching points on a regular basis)

2

The only impact on performance is that the window needs to keep more data around since it is receiving more data. Other than that it should work the same.