Package org.apache.hadoop.metrics2.util

Class SampleQuantiles

java.lang.Object

org.apache.hadoop.metrics2.util.SampleQuantiles

All Implemented Interfaces:

QuantileEstimator

@Private
public class SampleQuantiles
extends java.lang.Object
implements QuantileEstimator

Implementation of the Cormode, Korn, Muthukrishnan, and Srivastava algorithm for streaming calculation of targeted high-percentile epsilon-approximate quantiles. This is a generalization of the earlier work by Greenwald and Khanna (GK), which essentially allows different error bounds on the targeted quantiles, which allows for far more efficient calculation of high-percentiles. See: Cormode, Korn, Muthukrishnan, and Srivastava "Effective Computation of Biased Quantiles over Data Streams" in ICDE 2005 Greenwald and Khanna, "Space-efficient online computation of quantile summaries" in SIGMOD 2001

Constructor Summary

Constructors

Constructor	Description
SampleQuantiles(Quantile[] quantiles)

Method Summary

Modifier and Type	Method	Description
void	clear()	Resets the estimator, clearing out all previously inserted items
long	getCount()	Returns the number of items that the estimator has processed
int	getSampleCount()	Returns the number of samples kept by the estimator
void	insert(long v)	Add a new value from the stream.
java.util.Map<Quantile,java.lang.Long>	snapshot()	Get a snapshot of the current values of all the tracked quantiles.
java.lang.String	toString()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

SampleQuantiles

public SampleQuantiles(Quantile[] quantiles)

Method Detail

insert

public void insert(long v)

Add a new value from the stream.

Specified by:

insert in interface QuantileEstimator

Parameters:

v - v.

snapshot

public java.util.Map<Quantile,java.lang.Long> snapshot()

Get a snapshot of the current values of all the tracked quantiles.

Specified by:

snapshot in interface QuantileEstimator

Returns:

snapshot of the tracked quantiles. If no items are added to the estimator, returns null.

getCount

public long getCount()

Returns the number of items that the estimator has processed

Specified by:

getCount in interface QuantileEstimator

Returns:

count total number of items processed

getSampleCount

@VisibleForTesting
public int getSampleCount()

Returns the number of samples kept by the estimator

Returns:

count current number of samples

clear

public void clear()

Resets the estimator, clearing out all previously inserted items

Specified by:

clear in interface QuantileEstimator

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object