org.apache.flink.runtime.state

Class KeyGroupRangeAssignment

java.lang.Object

org.apache.flink.runtime.state.KeyGroupRangeAssignment

public final class KeyGroupRangeAssignment
extends Object

Field Summary

Fields

Modifier and Type	Field and Description
static int	DEFAULT_LOWER_BOUND_MAX_PARALLELISM The default lower bound for max parallelism if nothing was configured by the user.
static int	UPPER_BOUND_MAX_PARALLELISM The (inclusive) upper bound for max parallelism.

Method Summary

Modifier and Type	Method and Description
static int	assignKeyToParallelOperator(Object key, int maxParallelism, int parallelism) Assigns the given key to a parallel operator index.
static int	assignToKeyGroup(Object key, int maxParallelism) Assigns the given key to a key-group index.
static void	checkParallelismPreconditions(int parallelism)
static int	computeDefaultMaxParallelism(int operatorParallelism) Computes a default maximum parallelism from the operator parallelism.
static int	computeKeyGroupForKeyHash(int keyHash, int maxParallelism) Assigns the given key to a key-group index.
static KeyGroupRange	computeKeyGroupRangeForOperatorIndex(int maxParallelism, int parallelism, int operatorIndex) Computes the range of key-groups that are assigned to a given operator under the given parallelism and maximum parallelism.
static int	computeOperatorIndexForKeyGroup(int maxParallelism, int parallelism, int keyGroupId) Computes the index of the operator to which a key-group belongs under the given parallelism and maximum parallelism.

Methods inherited from class java.lang.Object

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

Field Detail

DEFAULT_LOWER_BOUND_MAX_PARALLELISM

public static final int DEFAULT_LOWER_BOUND_MAX_PARALLELISM

The default lower bound for max parallelism if nothing was configured by the user. We have this so allow users some degree of scale-up in case they forgot to configure maximum parallelism explicitly.

See Also:

Constant Field Values

UPPER_BOUND_MAX_PARALLELISM

public static final int UPPER_BOUND_MAX_PARALLELISM

The (inclusive) upper bound for max parallelism.

See Also:

Constant Field Values

Method Detail

assignKeyToParallelOperator

public static int assignKeyToParallelOperator(Object key,
                                              int maxParallelism,
                                              int parallelism)

Assigns the given key to a parallel operator index.

Parameters:

key - the key to assign

maxParallelism - the maximum supported parallelism, aka the number of key-groups.

parallelism - the current parallelism of the operator

Returns:

the index of the parallel operator to which the given key should be routed.

assignToKeyGroup

public static int assignToKeyGroup(Object key,
                                   int maxParallelism)

Assigns the given key to a key-group index.

Parameters:

key - the key to assign

maxParallelism - the maximum supported parallelism, aka the number of key-groups.

Returns:

the key-group to which the given key is assigned

computeKeyGroupForKeyHash

public static int computeKeyGroupForKeyHash(int keyHash,
                                            int maxParallelism)

Assigns the given key to a key-group index.

Parameters:

keyHash - the hash of the key to assign

maxParallelism - the maximum supported parallelism, aka the number of key-groups.

Returns:

the key-group to which the given key is assigned

computeKeyGroupRangeForOperatorIndex

public static KeyGroupRange computeKeyGroupRangeForOperatorIndex(int maxParallelism,
                                                                 int parallelism,
                                                                 int operatorIndex)

Computes the range of key-groups that are assigned to a given operator under the given parallelism and maximum parallelism.

IMPORTANT: maxParallelism must be <= Short.MAX_VALUE to avoid rounding problems in this method. If we ever want to go beyond this boundary, this method must perform arithmetic on long values.

Parameters:

maxParallelism - Maximal parallelism that the job was initially created with.

parallelism - The current parallelism under which the job runs. Must be <= maxParallelism.

operatorIndex - Id of a key-group. 0 <= keyGroupID < maxParallelism.

Returns:

the computed key-group range for the operator.

computeOperatorIndexForKeyGroup

public static int computeOperatorIndexForKeyGroup(int maxParallelism,
                                                  int parallelism,
                                                  int keyGroupId)

Computes the index of the operator to which a key-group belongs under the given parallelism and maximum parallelism.

IMPORTANT: maxParallelism must be <= Short.MAX_VALUE to avoid rounding problems in this method. If we ever want to go beyond this boundary, this method must perform arithmetic on long values.

Parameters:

maxParallelism - Maximal parallelism that the job was initially created with. 0 < parallelism <= maxParallelism <= Short.MAX_VALUE must hold.

parallelism - The current parallelism under which the job runs. Must be <= maxParallelism.

keyGroupId - Id of a key-group. 0 <= keyGroupID < maxParallelism.

Returns:

The index of the operator to which elements from the given key-group should be routed under the given parallelism and maxParallelism.

computeDefaultMaxParallelism

public static int computeDefaultMaxParallelism(int operatorParallelism)

Computes a default maximum parallelism from the operator parallelism. This is used in case the user has not explicitly configured a maximum parallelism to still allow a certain degree of scale-up.

Parameters:

operatorParallelism - the operator parallelism as basis for computation.

Returns:

the computed default maximum parallelism.

checkParallelismPreconditions

public static void checkParallelismPreconditions(int parallelism)