A real case, put together end to end: a DPS named “CPU average (Nov 2025 week)” selecting the
DataServer agent on host hex-perf20, with Metric type = AVG, a Greater than or equal to
10 value filter, Resample target points = 60, Reduce scope = One per series, and
Reduce function = Average. Its time period selector pins an absolute week rather than following
the portal’s current period.

1. Metric type selector
Metric type = AVG — every matched point’s value is the metric’s aggregate field. The metric name
regexp CPU:.* matches exactly two metrics on that agent, which List metrics confirms before
evaluating anything:

-
CPU:Processor Count— a near-constant metric:8, the number of CPU cores. -
CPU:Utilization % (process)— a genuinely fluctuating percentage, ranging from0to95over this period.
2. Values filtering
Greater than or equal to 10 keeps only raw points whose value is >= 10.
-
CPU:Processor Count’s value never reaches 10 (it stays at8for the whole period) — every point of this series is filtered out, leaving it with an empty points list. -
CPU:Utilization % (process)’s value regularly crosses 10 — the points above the threshold survive.
3. Reduce
One per series / Average computes one average per series directly from the filtered raw
points — resample hasn’t run yet at this point in the pipeline.
-
CPU:Processor Count—Averageof an empty points list is “no value” (not0), and a series with no value to report is dropped from a “One per series” result (see Reduce) — it does not appear at all, not even as a blank row. -
CPU:Utilization % (process)— a real averaged value, computed from every surviving raw point.
4. Resample
Resample target points = 60 is configured, but has no effect here: One per series already
collapsed each series down to a single value in step 3, so there’s nothing left for resample to
bucket (see Resample). Had this DPS instead used Reduce scope = One serie, the
60-point target would apply to the merged series that scope produces.
Result
Two metrics went in, one row comes out: CPU:Processor Count is absent, filtered out in step 2,
while CPU:Utilization % (process) shows its computed average.

Removing the value filter (or lowering the threshold below 8) makes CPU:Processor Count
reappear in the result, since its points would no longer be filtered out in step 2.