Aggregators

NameDescription
countNumber of samples during the period.
minMinimum value during the period.
maxMaximum value during the period.
sumSum of values during the period.
avgAverage value during the period.
percentile_999
percentile_995
percentile_99
percentile_95
percentile_90
percentile_75
percentile_50
Percentile(N) is a number which is greater than the N percentage of samples in the period.
For example, percentile_995 (99.5%) represents a number greater than 99.5% of all observations received during the period. The remaining 0.5% of all observations are greater than percentile_995.
Percentile argument can be any number between 0 and 100, for example: percentile_25.
The NIST (National Institute of Standards and Technology) method is used to calculate percentiles.
medianpercentile_50
standard_deviationStandard deviation of values during the period, square root of the variance.
firstFirst value received during the period.
lastLast value received during the period.
deltaDelta is computed as the difference between the last value in the period and the last value in the previous period. If no last value is present in the previous period, then delta is calculated as the difference between the last and first values during the current period. See also counter aggregator.
counterReturns sum of the differences between consecutive values in the period. If the difference between any two values is negative, the difference is replaced with the value itself. If the difference between values is always non-negative, the counter aggregator will return the same value as the delta aggregator. Example explaining difference between COUNTER and DELTA: View
wtavgAn average resulting from the multiplication of each component by a factor reflecting its importance.
Weight is assigned based on timestamps rather than index.
Data points with older timestamps contribute significantly less to the weighted mean than data points with more recent timestamps.
The weight of a sample is proportional to the current_time – first_time + 1 (in seconds), where current_time is the timestamp of the sample, and first_time is the timestamp of the first sample received during the period.
The weights are normalized so that their sum is equal to 1.
See calculation example below.
wavgA weighted average of samples received during the period.
The weight of a sample is proportional to the index of the sample in the time-ordered array of all samples during the period.
The weights are normalized so that their sum is equal to 1.
A weighted average is sum of the sample’s values multiplied by their weights.
See a calculation example below.
min_value_timeTime when the minimum value (min) occurred during the period.
max_value_timeTime when the maximum value (max) occurred during the period.
threshold_countNumber of threshold violations during the period.
A violation is a sequence of one or more successive data points exceeding the specified threshold.
Example: the following values are collected every 5 minutes: 1, 2, 5, 7, 1, 1
If the threshold is set to 4, then one sequence (5,7) exceeds the threshold.
threshold_durationTotal duration of threshold violations during the period, in milliseconds.
A violation is a sequence of one or more successive data points exceeding the specified threshold.
Example: the following values are collected every 5 minutes: 1, 2, 5, 7, 1, 1
If the threshold is set to 4, then one sequence (5,7) exceeded the threshold and its total duration based on linear interpolation is 550 seconds. If the threshold is 5, then duration is 850 seconds.
For periods at the beginning of the timespan, at the end of the timespan or after a missing period, the value at the beginning of the period is set to the first value, and the value at the end of the period is set to last value.
threshold_percentPercent of time during which the value was below threshold (no violation).
Formula: threshold_percent = 100 - threshold_duration / period_duration * 100

Interpolation Functions

If the period doesn’t contain any detailed values, it will not be included in the results. Aggregator behavior can be modified by specifying an interpolation function.

FunctionDescription
NONE (default)No interpolation. Periods without any raw values are excluded from results.
LINEARCalculate period value using linear interpolation between previous and next period values.
PREVIOUSSet value for the period based on the previous period’s value.
NEXTSet value for the period based on the next period’s value.
VALUESet value for the period to a specific number.

Interpolation Examples:

aggregate_interpolate

View in Chartlab

wtavg() – weighted time average.

Calculation: Weight = (sample.time - first.time)/(last.time - first.time + 1)

Aggregation Example

TimestampValueAggregators
2015-09-10 10:00:0011first = 11
delta = 0
2015-09-10 11:00:009min = 9 | min_value_time = 2015-09-10 11:00:00
delta = -2
2015-09-10 12:00:0014delta = 5
2015-09-10 13:00:0019delta = 5
2015-09-10 14:00:0021max = 21 | max_value_time = 2015-09-10 14:00:00
delta = 2
2015-09-10 15:00:0015delta = -6
2015-09-10 16:00:0014delta = -1
2015-09-10 17:00:0011delta = -3
2015-09-10 18:00:0016delta = 5
2015-09-10 19:00:0014delta = -2
2015-09-10 20:00:0013last = 13
delta = -1

Aggregators:
Threshold is set to 20
Period is set to 11 hours, so it includes with whole data set.

NOTE: Period starttime is inclusive and endtime is exclusive.
Meaning that startime = 2015-09-14 10:00:00 will include data points that occurred exactly at 10:00:00 and later.
Endtime = 2015-09-14 11:00:00 will include data points that occurred up to 10:59:59, excluding points that occurred at 11:00:00 and later.

count                  = 11
first                  = 11
last                   = 13
min                    = 9
min_value_time         = 2015-09-10 11:00:00
max                    = 21
max_value_time         = 2015-09-10 14:00:00
sum                    = 157
avg                    = 14.27
percentile_999         = 21
percentile_995         = 21
percentile_99          = 21
percentile_95          = 21
percentile_90          = 20.6
percentile_75          = 16
percentile_50 (median) = 14
standard_deviation     = 3.47
wavg                   = 14.47
wtavg                  = 14.51
threshold_count        = 1
threshold_duration     = 3600000
threshold_percent      = 88.9
Chart Lab Examples:

Box Chart illustrating the percentiles: view

Histogram illustrating the percentiles: view

Time Chart illustrating delta: view

Time Chart comparing counter and delta aggregators: view