T
- public interface Partitioner<T> extends Serializable
Multiple partitions are permitted. Only one partition is consider primary. A primary partition is the partition for an item in which the item is processed on behalf of itself. All other partitions are those partitions that require visibility to the a specific item for other items to reference. This approach supports nearest neighbor type queries. Consider that an item can only discover neighbors in its partition. However, the item can be discovered as a nearest neighbor in those partitions in which the item participates as a none primary.
Modifier and Type | Interface and Description |
---|---|
static class |
Partitioner.PartitionData
Represents a partition associated with a specific item.
|
static interface |
Partitioner.PartitionDataCallback |
Modifier and Type | Method and Description |
---|---|
List<Partitioner.PartitionData> |
getCubeIdentifiers(T entry) |
Collection<ParameterEnum<?>> |
getParameters() |
void |
initialize(org.apache.hadoop.mapreduce.JobContext context,
Class<?> scope) |
void |
partition(T entry,
Partitioner.PartitionDataCallback callback) |
void |
setup(PropertyManagement runTimeProperties,
Class<?> scope,
org.apache.hadoop.conf.Configuration configuration) |
void initialize(org.apache.hadoop.mapreduce.JobContext context, Class<?> scope) throws IOException
IOException
List<Partitioner.PartitionData> getCubeIdentifiers(T entry)
void partition(T entry, Partitioner.PartitionDataCallback callback) throws Exception
Exception
Collection<ParameterEnum<?>> getParameters()
void setup(PropertyManagement runTimeProperties, Class<?> scope, org.apache.hadoop.conf.Configuration configuration)
Copyright © 2013–2022. All rights reserved.