Reputation: 187
Struggling figuring out what exactly this thing means in DataDog:
avg:system.io.r-s{app:you} by {host}.rollup(avg,3600)
Why do we use first avg here in case we’ve already defined one datapoint that is avg of values within 3600. What the first avg does here?
Upvotes: 3
Views: 2535
Reputation: 1943
first avg
controls space aggregation. second one controls time aggregation.
if you have a dozen servers how should those servers be combined into a single line? that is what space aggregation decides. for system.io you coud imagine sum
could be more interesting than average sometimes. In this case because are you grouping by host
there is only 1 "group" per line, so the space aggregation does nothing.
when you zoom out far enough on a graph, you eventually aren't physically able to draw every single data point for every second measured. Each data point will represent a 5 minutes or 10 minute or 48 hour set of data. How do we decide how to combine all the seconds in a 5 minute window? that is what time aggregation decides. If you are zoomed in close enough (or the data is sparse enough) there is one one measurement per data point, and so time aggregation does nothing.
https://docs.datadoghq.com/dashboards/guide/query-to-the-graph/
Upvotes: 3