org.apache.druid.server.coordinator.balancer

Class CostBalancerStrategy

java.lang.Object

org.apache.druid.server.coordinator.balancer.CostBalancerStrategy

All Implemented Interfaces:

BalancerStrategy

Direct Known Subclasses:

CachingCostBalancerStrategy, DiskNormalizedCostBalancerStrategy

public class CostBalancerStrategy
extends Object
implements BalancerStrategy

Constructor Summary

Constructors

Constructor and Description
CostBalancerStrategy(com.google.common.util.concurrent.ListeningExecutorService exec)

Method Summary

Modifier and Type	Method and Description
static double	computeJointSegmentsCost(DataSegment segmentA, DataSegment segmentB) This defines the unnormalized cost function between two segments.
static double	computeJointSegmentsCost(DataSegment segment, Iterable<DataSegment> segmentSet)
protected double	computePlacementCost(DataSegment proposalSegment, ServerHolder server) Computes the cost of placing a segment on this server.
ServerHolder	findDestinationServerToMoveSegment(DataSegment segmentToMove, ServerHolder sourceServer, List<ServerHolder> serverHolders) Finds the best server from the list of destinationServers to load the segmentToMove, if it is moved from the sourceServer.
Iterator<ServerHolder>	findServersToDropSegment(DataSegment segmentToDrop, List<ServerHolder> serverHolders) Finds the best servers to drop the given segment.
Iterator<ServerHolder>	findServersToLoadSegment(DataSegment segmentToLoad, List<ServerHolder> serverHolders) Finds the best servers to load the given segment.
CoordinatorRunStats	getAndResetStats() Returns the stats collected by the strategy in the current run and resets the stats collector for the next run.
static double	intervalCost(double x1, double y0, double y1) Computes the joint cost of two intervals X = [x_0 = 0, x_1) and Y = [y_0, y_1) cost(X, Y) = \int_{x_0}^{x_1} \int_{y_0}^{y_1} e^{-\lambda |x-y|}dxdy $$ lambda = 1 in this particular implementation Other values of lambda can be calculated by multiplying inputs by lambda and multiplying the result by 1 / lambda ^ 2 Interval start and end are all relative to x_0.
static double	intervalCost(org.joda.time.Interval intervalA, org.joda.time.Interval intervalB)

Methods inherited from class java.lang.Object

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

Constructor Detail

CostBalancerStrategy

public CostBalancerStrategy(com.google.common.util.concurrent.ListeningExecutorService exec)

Method Detail

computeJointSegmentsCost

public static double computeJointSegmentsCost(DataSegment segment,
                                              Iterable<DataSegment> segmentSet)

computeJointSegmentsCost

public static double computeJointSegmentsCost(DataSegment segmentA,
                                              DataSegment segmentB)

This defines the unnormalized cost function between two segments. See https://github.com/apache/druid/pull/2972 for more details about the cost function. intervalCost: segments close together are more likely to be queried together multiplier: if two segments belong to the same data source, they are more likely to be involved in the same queries

Parameters:

segmentA - The first DataSegment.

segmentB - The second DataSegment.

Returns:

the joint cost of placing the two DataSegments together on one node.

intervalCost

public static double intervalCost(org.joda.time.Interval intervalA,
                                  org.joda.time.Interval intervalB)

intervalCost

public static double intervalCost(double x1,
                                  double y0,
                                  double y1)

Computes the joint cost of two intervals X = [x_0 = 0, x_1) and Y = [y_0, y_1) cost(X, Y) = \int_{x_0}^{x_1} \int_{y_0}^{y_1} e^{-\lambda |x-y|}dxdy $$ lambda = 1 in this particular implementation Other values of lambda can be calculated by multiplying inputs by lambda and multiplying the result by 1 / lambda ^ 2 Interval start and end are all relative to x_0. Therefore this function assumes x_0 = 0, x1 >= 0, and y1 > y0

Parameters:

x1 - end of interval X

y0 - start of interval Y

y1 - end o interval Y

Returns:

joint cost of X and Y

findServersToLoadSegment

public Iterator<ServerHolder> findServersToLoadSegment(DataSegment segmentToLoad,
                                                       List<ServerHolder> serverHolders)

Description copied from interface: BalancerStrategy

Finds the best servers to load the given segment. This method can be used both for placing the first copy of a segment in a tier or a replica of an already available segment.

Specified by:

findServersToLoadSegment in interface BalancerStrategy

Returns:

Iterator over the best servers (in order of preference) to load the segment.

findDestinationServerToMoveSegment

public ServerHolder findDestinationServerToMoveSegment(DataSegment segmentToMove,
                                                       ServerHolder sourceServer,
                                                       List<ServerHolder> serverHolders)

Description copied from interface: BalancerStrategy

Finds the best server from the list of destinationServers to load the segmentToMove, if it is moved from the sourceServer.

In order to avoid unnecessary moves when the segment is already optimally placed, include the sourceServer in the list of destinationServers.

Specified by:

findDestinationServerToMoveSegment in interface BalancerStrategy

Returns:

The server to move to, or null if the segment is already optimally placed.

findServersToDropSegment

public Iterator<ServerHolder> findServersToDropSegment(DataSegment segmentToDrop,
                                                       List<ServerHolder> serverHolders)

Description copied from interface: BalancerStrategy

Finds the best servers to drop the given segment.

Specified by:

findServersToDropSegment in interface BalancerStrategy

Returns:

Iterator over the servers (in order of preference) to drop the segment

getAndResetStats

public CoordinatorRunStats getAndResetStats()

Description copied from interface: BalancerStrategy

Returns the stats collected by the strategy in the current run and resets the stats collector for the next run.

Specified by:

getAndResetStats in interface BalancerStrategy

computePlacementCost

protected double computePlacementCost(DataSegment proposalSegment,
                                      ServerHolder server)

Computes the cost of placing a segment on this server.