Datadog functions

gistfile1.md

A short introduction to Datadog functions

You can apply functions to metric queries in the graph editor, as long as you use the JSON editor.

The general format is:

function(metric{scope} [by {filter}])

In case of binary operators (+, -, /, *), the format is:

metric{scope} [by {filter}] operator metric{scope} [by {filter}]

The filter part is optional.

Simple arithmetic

• query + query
• query - query
• query * query
• query / query

Derivation and integration

• dt(query): time delta between points
• diff(query): value delta between points
• derivative(query): 1st order derivative, diff / dt
• integral(query): sum(value * dt)
• cumsum(query): sum(value)
• rate(query): 1st order derivate that skips non-monotonically increasing values

Some syntactic sugar:

• derived(query): synonym for derivative
• per_second(query): synonym for rate
• per_minute(query): 60 * rate
• per_hour(query): 3600 * rate

Moving Averages

• avg_5min(query): average over past 5 minutes
• avg_10min(query): ditto for 10 minutes
• avg_30min(query): ...
• avg_hourly(query): ...
• avg_daily(query): ...

Exponentially-weighted averages using variable decay

(from pandas)

• ewma_1(query): decay = 1, same as average
• ewma_3(query): decay = 1/2
• ewma_5(query): decay = 1/3
• ewma_10(query): decay = 2/11
• ewma_20(query): decay = 2/21