Package io.trino.operator.output

Class SkewedPartitionRebalancer

java.lang.Object

io.trino.operator.output.SkewedPartitionRebalancer

@ThreadSafe
public class SkewedPartitionRebalancer
extends Object

Helps in distributing big or skewed partitions across available tasks to improve the performance of partitioned writes.

This rebalancer initialize a bunch of buckets for each task based on a given taskBucketCount and then tries to uniformly distribute partitions across those buckets. This helps to mitigate two problems: 1. Mitigate skewness across tasks. 2. Scale few big partitions across tasks even if there's no skewness among them. This will essentially speed the local scaling without impacting much overall resource utilization.

Example:

Before: 3 tasks, 3 buckets per task, and 2 skewed partitions Task1 Task2 Task3 Bucket1 (Part 1) Bucket1 (Part 2) Bucket1 Bucket2 Bucket2 Bucket2 Bucket3 Bucket3 Bucket3

After rebalancing: Task1 Task2 Task3 Bucket1 (Part 1) Bucket1 (Part 2) Bucket1 (Part 1) Bucket2 (Part 2) Bucket2 (Part 1) Bucket2 (Part 2) Bucket3 Bucket3 Bucket3

Method Summary

Modifier and Type	Method	Description
void	addDataProcessed(long dataSize)
void	addPartitionRowCount(int partition, long rowCount)
static boolean	checkCanScalePartitionsRemotely(Session session, int taskCount, PartitioningHandle partitioningHandle, NodePartitioningManager nodePartitioningManager)
static PartitionFunction	createPartitionFunction(Session session, NodePartitioningManager nodePartitioningManager, PartitioningScheme scheme, List<Type> partitionChannelTypes)
static SkewedPartitionRebalancer	createSkewedPartitionRebalancer(int partitionCount, int taskCount, int taskPartitionedWriterCount, long minPartitionDataProcessedRebalanceThreshold)
int	getTaskCount()
static int	getTaskCount(PartitioningScheme partitioningScheme)
int	getTaskId(int partitionId, long index)
void	rebalance()

Methods inherited from class java.lang.Object

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

Method Details

checkCanScalePartitionsRemotely

public static boolean checkCanScalePartitionsRemotely(Session session,
 int taskCount,
 PartitioningHandle partitioningHandle,
 NodePartitioningManager nodePartitioningManager)

createPartitionFunction

public static PartitionFunction createPartitionFunction(Session session,
 NodePartitioningManager nodePartitioningManager,
 PartitioningScheme scheme,
 List<Type> partitionChannelTypes)

createSkewedPartitionRebalancer

public static SkewedPartitionRebalancer createSkewedPartitionRebalancer(int partitionCount,
 int taskCount,
 int taskPartitionedWriterCount,
 long minPartitionDataProcessedRebalanceThreshold)

getTaskCount

public static int getTaskCount(PartitioningScheme partitioningScheme)

getTaskCount

public int getTaskCount()

getTaskId

public int getTaskId(int partitionId,
 long index)

addDataProcessed

public void addDataProcessed(long dataSize)

addPartitionRowCount

public void addPartitionRowCount(int partition,
 long rowCount)

rebalance

public void rebalance()