Wire up DIFFERENCE aggregate

[pauldix]

If the query is like this:

SELECT DIFFERENCE(value)
FROM cpu
WHERE time > now() - 4h

It will return a series with the difference between each value.

If the query is like this:

SELECT DIFFERENCE(MIN(value))
FROM cpu
WHERE time > now() - 4h
GROUP BY time(5m)

It will return the difference of the min value for each 5 minute window of time.

[Gibheer]

From tsdb/executor.go#L443 this seems to be implemented now, is that correct?

[jchauncey]

Any news on when this might come?

[Vebryn]

I don't see this issue on 9.5 changelog https://github.com/influxdb/influxdb/blob/master/CHANGELOG.md

[beckettsean]

@Vebryn it hasn't been completed, so it is not on the CHANGELOG.

[Vebryn]

Ok, thank you, let's wait and see

[superdump]

Looking forward to this.

[corylanou]

The query engine is undergoing heavy refactoring. Removing help wanted label for now.

[marcinpraczko]

Hi, Are there any ETA for this when it will be released?
Or what is other way of displaying bandwidth usage? We know that linux increases this number over and over: /sys/class/net/eth0/statistics/rx_bytes, so is there any way to do difference on InfluxDB level, or so far I have to do that differently by "collector"?

[pauldix]

What about non_negative_derivative?

[tothandor]

Could you please add a shorter alias for this function, like "diff" or "delta" instead of "non_negative_derivative"!

[jazdw]

Is there currently anyway to define different grouping functions for time and between series? So we can do mean_series(last("value") - first("value")).

[beckettsean]

@jazdw no, there are no composite functions yet. That specific case is an interesting feature request, can you open a new issue to track it?

[beckettsean]

As mentioned in my post to the mailing list we are experimenting with simplifying our open GitHub Issues. This feature request has been rolled into an aggregate issue for all function requests, so that we can close this issue until we are ready to work on it.

You may continue to make comments here. Closing the issue does not mean we are rejecting this idea.

[jsternberg]

I'm beginning to work on this.

[alcroito]

I know this is an old closed issue, but it still pops up when googling.
As requested by some previous commenters, the solution for computing bandwidth usage for a specific time period (or disk usage increase / decrease) is described in the final comments of #7076 and involves the usage of difference and cumulative_sum.