org.apache.flink.runtime.operators.hash

Class MutableHashTable<BT,PT>

java.lang.Object

org.apache.flink.runtime.operators.hash.MutableHashTable<BT,PT>

Type Parameters:

BT - The type of records from the build side that are stored in the hash table.

PT - The type of records from the probe side that are stored in the hash table.

All Implemented Interfaces:

MemorySegmentSource

Direct Known Subclasses:

ReOpenableMutableHashTable

public class MutableHashTable<BT,PT>
extends Object
implements MemorySegmentSource

An implementation of a Hybrid Hash Join. The join starts operating in memory and gradually starts spilling contents to disk, when the memory is not sufficient. It does not need to know a priori how large the input will be.

The design of this class follows on many parts the design presented in "Hash joins and hash teams in Microsoft SQL Server", by Goetz Graefe et al. In its current state, the implementation lacks features like dynamic role reversal, partition tuning, or histogram guided partitioning.

The layout of the buckets inside a memory segment is as follows:

 +----------------------------- Bucket x ----------------------------
 |Partition (1 byte) | Status (1 byte) | element count (2 bytes) |
 | next-bucket-in-chain-pointer (8 bytes) | reserved (4 bytes) |
 |
 |hashCode 1 (4 bytes) | hashCode 2 (4 bytes) | hashCode 3 (4 bytes) |
 | ... hashCode n-1 (4 bytes) | hashCode n (4 bytes)
 |
 |pointer 1 (8 bytes) | pointer 2 (8 bytes) | pointer 3 (8 bytes) |
 | ... pointer n-1 (8 bytes) | pointer n (8 bytes)
 |
 +---------------------------- Bucket x + 1--------------------------
 |Partition (1 byte) | Status (1 byte) | element count (2 bytes) |
 | next-bucket-in-chain-pointer (8 bytes) | reserved (4 bytes) |
 |
 |hashCode 1 (4 bytes) | hashCode 2 (4 bytes) | hashCode 3 (4 bytes) |
 | ... hashCode n-1 (4 bytes) | hashCode n (4 bytes)
 |
 |pointer 1 (8 bytes) | pointer 2 (8 bytes) | pointer 3 (8 bytes) |
 | ... pointer n-1 (8 bytes) | pointer n (8 bytes)
 +-------------------------------------------------------------------
 | ...
 |
 

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	MutableHashTable.HashBucketIterator<BT,PT>
static class	MutableHashTable.ProbeIterator<PT>

Field Summary

Fields

Modifier and Type	Field and Description
protected List<MemorySegment>	availableMemory The free memory segments currently available to the hash join.
protected MemorySegment[]	buckets The array of memory segments that contain the buckets which form the actual hash-table of hash-codes and pointers to the elements.
protected int	bucketsPerSegmentBits The number of bits that describe the position of a bucket in a memory segment.
protected int	bucketsPerSegmentMask The number of hash table buckets in a single memory segment - 1.
protected TypeComparator<BT>	buildSideComparator The utilities to hash and compare the build side data types.
protected TypeSerializer<BT>	buildSideSerializer The utilities to serialize the build side data types.
protected AtomicBoolean	closed Flag indicating that the closing logic has been invoked.
protected Channel.Enumerator	currentEnumerator The channel enumerator that is used while processing the current partition to create channels for the spill partitions it requires.
protected int	currentRecursionDepth The recursion depth of the partition that is currently processed.
protected boolean	furtherPartitioning
protected IOManager	ioManager The I/O manager used to instantiate writers for the spilled partitions.
protected boolean	keepBuildSidePartitions If true, build side partitions are kept for multiple probe steps.
protected int	numBuckets The number of buckets in the current table.
protected ArrayList<HashPartition<BT,PT>>	partitionsBeingBuilt The partitions that are built by processing the current partition.
protected MutableHashTable.ProbeIterator<PT>	probeIterator Iterator over the elements from the probe side.
protected TypeSerializer<PT>	probeSideSerializer The utilities to serialize the probe side data types.
protected int	segmentSize The size of the segments used by the hash join buckets.
protected LinkedBlockingQueue<MemorySegment>	writeBehindBuffers The queue of buffers that can be used for write-behind.
protected int	writeBehindBuffersAvailable The number of buffers in the write behind queue that are actually not write behind buffers, but regular buffers that only have not yet returned.

Constructor Summary

Constructors

Constructor and Description
MutableHashTable(TypeSerializer<BT> buildSideSerializer, TypeSerializer<PT> probeSideSerializer, TypeComparator<BT> buildSideComparator, TypeComparator<PT> probeSideComparator, TypePairComparator<PT,BT> comparator, List<MemorySegment> memorySegments, IOManager ioManager)
MutableHashTable(TypeSerializer<BT> buildSideSerializer, TypeSerializer<PT> probeSideSerializer, TypeComparator<BT> buildSideComparator, TypeComparator<PT> probeSideComparator, TypePairComparator<PT,BT> comparator, List<MemorySegment> memorySegments, IOManager ioManager, int avgRecordLen)

Method Summary

Methods

Modifier and Type	Method and Description
void	abort()
static byte	assignPartition(int bucket, byte numPartitions) Assigns a partition to a bucket.
protected void	buildInitialTable(MutableObjectIterator<BT> input)
protected void	buildTableFromSpilledPartition(HashPartition<BT,PT> p)
protected void	clearPartitions() This method clears all partitions currently residing (partially) in memory.
void	close() Closes the hash table.
protected void	createPartitions(int numPartitions, int recursionLevel)
MutableHashTable.HashBucketIterator<BT,PT>	getBuildSideIterator()
PT	getCurrentProbeRecord()
List<MemorySegment>	getFreedMemory()
static int	getInitialTableSize(int numBuffers, int bufferSize, int numPartitions, int recordLenBytes)
MutableHashTable.HashBucketIterator<BT,PT>	getMatchesFor(PT record)
protected HashPartition<BT,PT>	getNewInMemoryPartition(int number, int recursionLevel) Returns a new inMemoryPartition object.
static int	getNumWriteBehindBuffers(int numBuffers) Determines the number of buffers to be used for asynchronous write behind.
MutableObjectIterator<BT>	getPartitionEntryIterator()
static int	getPartitioningFanOutNoEstimates(int numBuffers) Gets the number of partitions to be used for an initial hash-table, when no estimates are available.
TypeComparator<PT>	getProbeSideComparator()
static int	hash(int code, int level) This function hashes an integer value.
protected void	initTable(int numBuckets, byte numPartitions)
protected void	insertIntoTable(BT record, int hashCode)
boolean	nextRecord()
MemorySegment	nextSegment()
void	open(MutableObjectIterator<BT> buildSide, MutableObjectIterator<PT> probeSide) Opens the hash join.
protected boolean	prepareNextPartition()
protected boolean	processProbeIter()
protected void	releaseTable() Releases the table (the array of buckets) and returns the occupied memory segments to the list of free segments.
protected int	spillPartition() Selects a partition and spills it.

Methods inherited from class java.lang.Object

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

Field Detail

buildSideSerializer

protected final TypeSerializer<BT> buildSideSerializer

The utilities to serialize the build side data types.

probeSideSerializer

protected final TypeSerializer<PT> probeSideSerializer

The utilities to serialize the probe side data types.

buildSideComparator

protected final TypeComparator<BT> buildSideComparator

The utilities to hash and compare the build side data types.

availableMemory

protected final List<MemorySegment> availableMemory

The free memory segments currently available to the hash join.

writeBehindBuffers

protected final LinkedBlockingQueue<MemorySegment> writeBehindBuffers

The queue of buffers that can be used for write-behind. Any buffer that is written asynchronously to disk is returned through this queue. hence, it may sometimes contain more

ioManager

protected final IOManager ioManager

The I/O manager used to instantiate writers for the spilled partitions.

segmentSize

protected final int segmentSize

The size of the segments used by the hash join buckets. All segments must be of equal size to ease offset computations.

bucketsPerSegmentMask

protected final int bucketsPerSegmentMask

The number of hash table buckets in a single memory segment - 1. Because memory segments can be comparatively large, we fit multiple buckets into one memory segment. This variable is a mask that is 1 in the lower bits that define the number of a bucket in a segment.

bucketsPerSegmentBits

protected final int bucketsPerSegmentBits

The number of bits that describe the position of a bucket in a memory segment. Computed as log2(bucketsPerSegment).

partitionsBeingBuilt

protected final ArrayList<HashPartition<BT,PT>> partitionsBeingBuilt

The partitions that are built by processing the current partition.

probeIterator

protected MutableHashTable.ProbeIterator<PT> probeIterator

Iterator over the elements from the probe side.

currentEnumerator

protected Channel.Enumerator currentEnumerator

The channel enumerator that is used while processing the current partition to create channels for the spill partitions it requires.

buckets

protected MemorySegment[] buckets

The array of memory segments that contain the buckets which form the actual hash-table of hash-codes and pointers to the elements.

numBuckets

protected int numBuckets

The number of buckets in the current table. The bucket array is not necessarily fully used, when not all buckets that would fit into the last segment are actually used.

writeBehindBuffersAvailable

protected int writeBehindBuffersAvailable

The number of buffers in the write behind queue that are actually not write behind buffers, but regular buffers that only have not yet returned. This is part of an optimization that the spilling code needs not wait until the partition is completely spilled before proceeding.

currentRecursionDepth

protected int currentRecursionDepth

The recursion depth of the partition that is currently processed. The initial table has a recursion depth of 0. Partitions spilled from a table that is built for a partition with recursion depth n have a recursion depth of n+1.

closed

protected AtomicBoolean closed

Flag indicating that the closing logic has been invoked.

keepBuildSidePartitions

protected boolean keepBuildSidePartitions

If true, build side partitions are kept for multiple probe steps.

furtherPartitioning

protected boolean furtherPartitioning

Constructor Detail

MutableHashTable

public MutableHashTable(TypeSerializer<BT> buildSideSerializer,
                TypeSerializer<PT> probeSideSerializer,
                TypeComparator<BT> buildSideComparator,
                TypeComparator<PT> probeSideComparator,
                TypePairComparator<PT,BT> comparator,
                List<MemorySegment> memorySegments,
                IOManager ioManager)

MutableHashTable

public MutableHashTable(TypeSerializer<BT> buildSideSerializer,
                TypeSerializer<PT> probeSideSerializer,
                TypeComparator<BT> buildSideComparator,
                TypeComparator<PT> probeSideComparator,
                TypePairComparator<PT,BT> comparator,
                List<MemorySegment> memorySegments,
                IOManager ioManager,
                int avgRecordLen)

Method Detail

open

public void open(MutableObjectIterator<BT> buildSide,
        MutableObjectIterator<PT> probeSide)
          throws IOException

Opens the hash join. This method reads the build-side input and constructs the initial hash table, gradually spilling partitions that do not fit into memory.

Throws:

IOException - Thrown, if an I/O problem occurs while spilling a partition.

processProbeIter

protected boolean processProbeIter()
                            throws IOException

Throws:

IOException

prepareNextPartition

protected boolean prepareNextPartition()
                                throws IOException

Throws:

IOException

nextRecord

public boolean nextRecord()
                   throws IOException

Returns:

Throws:

IOException

getMatchesFor

public MutableHashTable.HashBucketIterator<BT,PT> getMatchesFor(PT record)
                                                         throws IOException

Throws:

IOException

getCurrentProbeRecord

public PT getCurrentProbeRecord()

Returns:

getBuildSideIterator

public MutableHashTable.HashBucketIterator<BT,PT> getBuildSideIterator()

Returns:

getPartitionEntryIterator

public MutableObjectIterator<BT> getPartitionEntryIterator()

close

public void close()

Closes the hash table. This effectively releases all internal structures and closes all open files and removes them. The call to this method is valid both as a cleanup after the complete inputs were properly processed, and as an cancellation call, which cleans up all resources that are currently held by the hash join.

abort

public void abort()

getFreedMemory

public List<MemorySegment> getFreedMemory()

buildInitialTable

protected void buildInitialTable(MutableObjectIterator<BT> input)
                          throws IOException

Parameters:

input -

Throws:

IOException

buildTableFromSpilledPartition

protected void buildTableFromSpilledPartition(HashPartition<BT,PT> p)
                                       throws IOException

Parameters:

p -

Throws:

IOException

insertIntoTable

protected final void insertIntoTable(BT record,
                   int hashCode)
                              throws IOException

Parameters:

record -

hashCode -

Throws:

IOException

getNewInMemoryPartition

protected HashPartition<BT,PT> getNewInMemoryPartition(int number,
                                           int recursionLevel)

Returns a new inMemoryPartition object. This is required as a plug for ReOpenableMutableHashTable.

createPartitions

protected void createPartitions(int numPartitions,
                    int recursionLevel)

Parameters:

numPartitions -

clearPartitions

protected void clearPartitions()

This method clears all partitions currently residing (partially) in memory. It releases all memory and deletes all spilled partitions.

This method is intended for a hard cleanup in the case that the join is aborted.

initTable

protected void initTable(int numBuckets,
             byte numPartitions)

Parameters:

numBuckets -

numPartitions -

releaseTable

protected void releaseTable()

Releases the table (the array of buckets) and returns the occupied memory segments to the list of free segments.

spillPartition

protected int spillPartition()
                      throws IOException

Selects a partition and spills it. The number of the spilled partition is returned.

Returns:

The number of the spilled partition.

Throws:

IOException

nextSegment

public MemorySegment nextSegment()

Specified by:

nextSegment in interface MemorySegmentSource

getNumWriteBehindBuffers

public static final int getNumWriteBehindBuffers(int numBuffers)

Determines the number of buffers to be used for asynchronous write behind. It is currently computed as the logarithm of the number of buffers to the base 4, rounded up, minus 2. The upper limit for the number of write behind buffers is however set to six.

Parameters:

numBuffers - The number of available buffers.

Returns:

The number

getPartitioningFanOutNoEstimates

public static final int getPartitioningFanOutNoEstimates(int numBuffers)

Gets the number of partitions to be used for an initial hash-table, when no estimates are available.

The current logic makes sure that there are always between 10 and 127 partitions, and close to 0.1 of the number of buffers.

Parameters:

numBuffers - The number of buffers available.

Returns:

The number of partitions to use.

getInitialTableSize

public static final int getInitialTableSize(int numBuffers,
                      int bufferSize,
                      int numPartitions,
                      int recordLenBytes)

assignPartition

public static final byte assignPartition(int bucket,
                   byte numPartitions)

Assigns a partition to a bucket.

Parameters:

bucket -

numPartitions -

Returns:

The hash code for the integer.

hash

public static final int hash(int code,
       int level)

This function hashes an integer value. It is adapted from Bob Jenkins' website http://www.burtleburtle.net/bob/hash/integer.html. The hash function has the full avalanche property, meaning that every bit of the value to be hashed affects every bit of the hash value.

Parameters:

code - The integer to be hashed.

Returns:

The hash code for the integer.

getProbeSideComparator

public TypeComparator<PT> getProbeSideComparator()