simpledb.parallel

Class PartitionFunction<K,V>

java.lang.Object

simpledb.parallel.PartitionFunction<K,V>

All Implemented Interfaces:

java.io.Serializable

Direct Known Subclasses:

SingleFieldHashPartitionFunction

public abstract class PartitionFunction<K,V>
extends java.lang.Object
implements java.io.Serializable

The ShuffleProducer class uses an instance of the PartitionFunction class to decide which worker a tuple should be routed to. Typically, the ShuffleProducer class invokes partition on every tuple it generates.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
(package private) java.util.HashMap<K,V>	attributes
(package private) int	numPartition
private static long	serialVersionUID

Constructor Summary

Constructors

Constructor and Description
PartitionFunction(int numPartition) Each partition function implementation must has a Class(int) style constructor

Method Summary

Methods

Modifier and Type	Method and Description
V	getAttribute(K attribute)
int	numPartition()
abstract int	partition(Tuple t, TupleDesc td) Given an input tuple t, determine which partition to route it to.
void	setAttribute(K attribute, V value) A concrete implementation of a partition function may need some information to help it decide the tuple partitions.

Methods inherited from class java.lang.Object

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

Field Detail

serialVersionUID

private static final long serialVersionUID

See Also:

Constant Field Values

attributes

java.util.HashMap<K,V> attributes

numPartition

int numPartition

Constructor Detail

PartitionFunction

public PartitionFunction(int numPartition)

Each partition function implementation must has a Class(int) style constructor

Method Detail

setAttribute

public void setAttribute(K attribute,
                V value)

A concrete implementation of a partition function may need some information to help it decide the tuple partitions.

getAttribute

public V getAttribute(K attribute)

numPartition

public int numPartition()

partition

public abstract int partition(Tuple t,
            TupleDesc td)

Given an input tuple t, determine which partition to route it to. Note: TupleDesc td is explicitly required even though the Tuple t includes a TupleDesc (obtained by calling t.getTupleDesc()) since field names might be absent from t.getTupleDesc(), and the PartitionFunction might require field names.

Parameters:

t - the input tuple to route.

td - the tuple descriptor of the input tuple. Must have non-null names for those attributes that are used to compute the worker to route to.

Returns:

the worker to send the tuple to.