Package io.confluent.kafka.replication.push.metrics

Class HdrHistogram

java.lang.Object

io.confluent.kafka.replication.push.metrics.HdrHistogram

public final class HdrHistogram
extends Object

A wrapper on top of the HdrHistogram API. It handles writing to the histogram by delegating to an internal ValueRecorder implementation, and reading from the histogram by efficiently implementing the retrieval of up-to-date histogram data (the exact implementation of these varies depending on the HdrHistogram.ThreadSafetyMode with which the provider is initialized).

Note that all APIs expect a timestamp which is used by the histogram to discard decaying data and determine when the snapshot from which the histogram metrics are calculated should be refreshed.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static enum	HdrHistogram.ThreadSafetyMode	Describes the desired thread safety guarantees of the internally maintained HdrHistogram ValueRecorder implementation.

Constructor Summary

Constructors

Constructor	Description
HdrHistogram(HdrHistogram.ThreadSafetyMode mode, long highestTrackableValue, int numberOfSignificantValueDigits)

Method Summary

Modifier and Type	Method	Description
long	count(long now)
long	max(long now)
double	mean(long now)
double	measurePercentile(long now, double percentile)	Reads percentile data from the histogram.
long	min(long now)
void	record(long value)	Writes to the histogram.
double	stdDeviation(long now)

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

HdrHistogram

public HdrHistogram(HdrHistogram.ThreadSafetyMode mode,
 long highestTrackableValue,
 int numberOfSignificantValueDigits)

Method Details

record

public void record(long value)

Writes to the histogram. Will throw ArrayIndexOutOfBoundsException if the histogram's highestTrackableValue is lower than the value being recorded.

Parameters:

value - The value to be recorded. Cannot be negative.

count

public long count(long now)

Parameters:

now - An externally provided timestamp expected to be in milliseconds.

Returns:

The total number of updates recorded by the histogram, i.e. the number of times record(long) has been called.

max

public long max(long now)

Parameters:

now - An externally provided timestamp expected to be in milliseconds.

Returns:

The maximum value recorded by the histogram.

mean

public double mean(long now)

Parameters:

now - An externally provided timestamp expected to be in milliseconds.

Returns:

The mean/average of all the values recorded by the histogram.

min

public long min(long now)

Parameters:

now - An externally provided timestamp expected to be in milliseconds.

Returns:

The minimum value recorded by the histogram.

stdDeviation

public double stdDeviation(long now)

Parameters:

now - An externally provided timestamp expected to be in milliseconds.

Returns:

The population standard deviation of all the values recorded by the histogram.

measurePercentile

public double measurePercentile(long now,
 double percentile)

Reads percentile data from the histogram.

Parameters:

now - An externally provided timestamp expected to be in milliseconds.

percentile - The percentile for which a value is going to be retrieved. Expected to be between 0.0 and 100.0.

Returns:

The histogram value for the given percentile.