#include <parallelHashJoin.h>

Inheritance diagram for ParallelHashJoin:

[legend]

Collaboration diagram for ParallelHashJoin:

[legend]

Public Member Functions
	ParallelHashJoin (FileId outer_id, FileId inner_id, FileId result_id, std::vector< FieldId > outer_fields, std::vector< FieldId > inner_fields, std::uint32_t num_buckets, std::uint32_t num_threads, std::string temp_path, Catalog catalog, BufferManager buf_mgr, FileManager *file_mgr)
	Constructor for the join operater using a hash join algorithm. Sets up the state for a single join operation using the specified state.

	~ParallelHashJoin ()
	Destructor for hashjoin class.

void	runOperation ()
	Performs the join operation using the hash join alogrithm.

Public Member Functions inherited from HashJoin
	HashJoin (FileId outer_id, FileId inner_id, FileId result_id, std::vector< FieldId > outer_fields, std::vector< FieldId > inner_fields, std::uint32_t num_buckets, std::string temp_path, Catalog catalog, BufferManager buf_mgr, FileManager *file_mgr)
	Constructor for the join operater using a hash join algorithm. Sets up the state for a single join operation using the specified state.

	~HashJoin ()
	Destructor for hashjoin class.

Public Member Functions inherited from Join
	Join (FileId outer_id, FileId inner_id, FileId result_id, std::vector< FieldId > outer_fields, std::vector< FieldId > inner_fields, Catalog *catalog)
	Constructor for Join operation. Join subclasses use this constructor.

	~Join ()
	Destructor for the Join Operation.

Public Member Functions inherited from Operation
	Operation (FileId result_id, Catalog *catalog)
	Constructor for the Operation class. Because Operation is an abstract class, an object cannot be created, but this constructor is used by derived classes.

virtual	~Operation ()
	Destructor for the Operation class. Cleans up dynamic memory in result state.

Private Member Functions
void	_firstHash (std::uint32_t tid, bool is_outer)
	Performs the initial hashing of each relation for the first step of hash join.

void	_secondHash (std::uint32_t tid, std::uint32_t n_threads)
	Performs the second hashing of each relation for the second step of hash join.

RecordId	_parallelPart1 (Record record, bool is_outer, BlockHeapFileScanner scanner)
	Performs the main looping functionality for first hash.

void	_barrier (std::uint32_t n_threads)
	Helper function for _parallelRun which has threads wait until all threads have called this function.

void	_parallelRun (std::uint32_t tid)
	Performs the parallel functionality of parallel run, to be called by all threads.

Private Attributes
std::uint32_t	num_threads

std::uint32_t	result_num

std::mutex *	mtx_table

std::mutex	mtx

std::condition_variable	cv

std::uint32_t	tcount

std::exception_ptr	thd_exc_ptr

std::mutex	exc_mtx

std::vector< std::thread >	threads

Additional Inherited Members
Protected Member Functions inherited from HashJoin
std::uint32_t	hash1 (Record *rec, bool is_outer)
	Performs the first hash function on the inputted record.

void	_createTempFiles ()
	Performs the set up phase of hash join, creating temporary hashed files to be used by hash join algorithm.

void	_firstHash (bool is_outer)
	Performs the initial hashing of each relation for the first step of hash join.

void	_secondHash ()
	Performs the second hashing of each relation for the second step of hash join.

RecordId	_part1 (Record record, bool is_outer, BlockHeapFileScanner scanner)
	Performs the main looping functionality for first hash.

void	cleanup ()
	Function that cleans up state and deletes all allocated memory. It will be called once by one thread when any function throws an error.

Protected Member Functions inherited from Operation
void	_initState (FileId file_id, std::vector< FieldId > fields, fileState *state)
	Performs the file and temporary record setup for relational operators.

void	_delState (fileState *file_state)
	Deletes objects created in relop structs.

Protected Attributes inherited from HashJoin
BufferManager *	buf_mgr

FileManager *	file_mgr

std::vector< HeapFile * >	outer_partitions

std::vector< HeapFile * >	inner_partitions

std::uint32_t	num_buckets

std::vector< std::pair< HeapPage *, PageId > >	hash_table

std::string	temp_path
	Holds the path to which the temp files should be saved. "/local/" is recommended for performance.

std::uint32_t	result_num

Protected Attributes inherited from Join
fileState	outer

fileState	inner

std::vector< FieldId >	outer_fields

std::vector< FieldId >	inner_fields

Protected Attributes inherited from Operation
fileState	result_state

Catalog *	catalog

Detailed Description

SwatDB ParallelHashJoin Class. The interface to relational operators layer of the system: manages relational operators.

Constructor & Destructor Documentation

◆ ParallelHashJoin()

ParallelHashJoin::ParallelHashJoin	(	FileId	outer_id,
		FileId	inner_id,
		FileId	result_id,
		std::vector< FieldId >	outer_fields,
		std::vector< FieldId >	inner_fields,
		std::uint32_t	num_buckets,
		std::uint32_t	num_threads,
		std::string	temp_path,
		Catalog *	catalog,
		BufferManager *	buf_mgr,
		FileManager *	file_mgr
	)

Constructor for the join operater using a hash join algorithm. Sets up the state for a single join operation using the specified state.

Precondition: The types of the fields being joined on have the same type.

Postcondition: The set up for the hash join is complete.

Parameters

outer_id.	FileId of the outer relation file.
inner_id.	FileId of the inner relation file.
result_id.	FileId of the result relation file.
outer_field_ids.	Vector of FieldIds corresponding to the join field in outer_rel
inner_field_ids.	Vector of FieldIds corresponding to the join field in inner_rel
num_buckets.	Amount of partitions for the hashing
num_threads.	Amount of threads for the computation
catalog.	Pointer to the catalog of the Swatdb object
buf_mgr.	Pointer to the buffer manager of the dbms
temp_path.	filename for temporary files

Member Function Documentation

◆ _barrier()

void ParallelHashJoin::_barrier ( std::uint32_t n_threads )

private

Helper function for _parallelRun which has threads wait until all threads have called this function.

Postcondition: All threads have reached this function.

◆ _firstHash()

void ParallelHashJoin::_firstHash	(	std::uint32_t	tid,
		bool	is_outer
	)

private

Performs the initial hashing of each relation for the first step of hash join.

Parameters

tid.	Thread id of the thread performing this specific round of \ computation.
is_outer.	Boolean indicator of which relation is being hashed

◆ _parallelPart1()

RecordId ParallelHashJoin::_parallelPart1	(	Record *	record,
		bool	is_outer,
		BlockHeapFileScanner *	scanner
	)

private

Performs the main looping functionality for first hash.

Parameters

record.	Pointer to the record object which holds the current rec
is_outer.	Boolean indicator of current relation
scanner.	Scanner object which is iterating over the relation

Returns: rid. Record id

◆ _parallelRun()

void ParallelHashJoin::_parallelRun ( std::uint32_t tid )

private

Performs the parallel functionality of parallel run, to be called by all threads.

Parameters

tid.	ThreadId of current thread

Postcondition: Parallel functionality of parallel run is complete and threads can be joined.

◆ runOperation()

void ParallelHashJoin::runOperation ( )

virtual

Performs the join operation using the hash join alogrithm.

Precondition: Join state correctly initialized

Postcondition: HeapFile contains all matching results. The object should be destroyed, state is invalid.

Reimplemented from HashJoin.

Member Data Documentation

◆ cv

std::condition_variable ParallelHashJoin::cv

private

Hashjoin condition variable to implement barrier synchronization

◆ exc_mtx

std::mutex ParallelHashJoin::exc_mtx

private

Exception mutex. Helps keep track of whether the exception ptr has been set for the threaded portion

◆ mtx

std::mutex ParallelHashJoin::mtx

private

Hashjoin mutex

◆ mtx_table

std::mutex* ParallelHashJoin::mtx_table

private

Stores an array of mutexs which corresponds to the hash table

◆ num_threads

std::uint32_t ParallelHashJoin::num_threads

private

The number of threads the relations are being hashed into.

◆ tcount

std::uint32_t ParallelHashJoin::tcount

private

counter to implement barrier synchronziation with cv

◆ thd_exc_ptr

std::exception_ptr ParallelHashJoin::thd_exc_ptr

private

Exception pointer to catch exceptions in threads After all threads are joined, check this value to rethrow exception

◆ threads

std::vector<std::thread> ParallelHashJoin::threads

private

Stores an array of threads which correspond to hash table

The documentation for this class was generated from the following file:

parallelHashJoin.h

Public Member Functions

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ ParallelHashJoin()

Member Function Documentation

◆ _barrier()

◆ _firstHash()

◆ _parallelPart1()

◆ _parallelRun()

◆ runOperation()

Member Data Documentation

◆ cv

◆ exc_mtx

◆ mtx

◆ mtx_table

◆ num_threads

◆ tcount

◆ thd_exc_ptr

◆ threads