Hi Dojo,
I would like to clarify how the 5-minute average is calculated for trending, specifically during an element timeout state.
If an element enters a timeout state one or more times within a 5-minute interval, how is the 5-minute average calculated during that period?
Specifically:
- Is the timeout state treated as a value of 0, which would reduce the overall average?
- Are timeout period excluded from the calculation altogether?
- Or is another calculation method used?
I would appreciate it if you could explain how the 5-minute average is calculated when timeout periods occur within the averaging interval.
Appreciate your assistance and Thank you
Hi Nur,
DataMiner uses a time-weighted average for 5-minute trending and not a simple mean. Each value is weighted by how long it was active within the window:
Average = Σ(value_i × duration_i) / Σ(duration_i)
During Element Timeout
When an element enters timeout, DataMiner writes special status markers to the trend database:
- -4 => element going into timeout
- -5 => element coming out of timeout
The behavior is:
- ❌ Timeout is NOT treated as a value of 0; no zero is injected into the calculation.
- ✅ Timeout period is excluded; the averaging window is written prematurely when timeout begins (using only data collected up to that point). When the element recovers, a new window starts fresh.
- The timeout interval appears as a gap in the average trend graph.
Summary Table
Scenario Effect on 5-min Average
Timeout occurs mid-window Window written early; gap during timeout
Timeout period itself Excluded (not treated as 0)
Recovery from timeout New 5-min window starts fresh
Exception values mixed with valid values Exceptions ignored; only valid values averaged
Hope this helps.
Regards,
Paulo