org.apache.flink.api.common.distributions

Interface DataDistribution

All Superinterfaces:

IOReadableWritable, Serializable

@PublicEvolving
public interface DataDistribution
extends IOReadableWritable, Serializable

Method Summary

Modifier and Type	Method and Description
Object[]	getBucketBoundary(int bucketNum, int totalNumBuckets) Returns the i'th bucket's upper bound, given that the distribution is to be split into totalBuckets buckets.
TypeInformation[]	getKeyTypes() Gets the type of the key by which the dataSet is partitioned.
int	getNumberOfFields() The number of fields in the (composite) key.

Methods inherited from interface org.apache.flink.core.io.IOReadableWritable

read, write

Method Detail

getBucketBoundary

Object[] getBucketBoundary(int bucketNum,
                           int totalNumBuckets)

Returns the i'th bucket's upper bound, given that the distribution is to be split into totalBuckets buckets.

Assuming n buckets, let B_i be the result from calling getBucketBoundary(i, n), then the distribution will partition the data domain in the following fashion:

 (-inf, B_1] (B_1, B_2] ... (B_n-2, B_n-1] (B_n-1, inf)
 

Note: The last bucket's upper bound is actually discarded by many algorithms. The last bucket is assumed to hold all values v such that v > getBucketBoundary(n-1, n), where n is the number of buckets.

Parameters:

bucketNum - The number of the bucket for which to get the upper bound.

totalNumBuckets - The number of buckets to split the data into.

Returns:

A record whose values act as bucket boundaries for the specified bucket.

getNumberOfFields

int getNumberOfFields()

The number of fields in the (composite) key. This determines how many fields in the records define the bucket. The number of fields must be the size of the array returned by the function getBucketBoundary(int, int).

Returns:

The number of fields in the (composite) key.

getKeyTypes

TypeInformation[] getKeyTypes()

Gets the type of the key by which the dataSet is partitioned.

Returns:

The type of the key by which the dataSet is partitioned.