New Concepts for Parallel Object-Relational Query Processing Book

Ch. 1	Introduction
	ORDBMS: The Next Great Wave	1
	Extensible DBMS	2
Ch. 2	Background on User-Defined Routines
	User-Defined Routines	5
	Definition, Implementation, and Execution of New UDR	6
	User-Defined Scalar Functions	7
	User-Defined Aggregate Functions	9
	User-Defined Table Functions	10
	User-Defined Functions and Large Objects	11
	Comparison with Stored Procedures	12
	Optimization of Queries with UDF	12
Ch. 3	Parallel Processing of User-Defined Functions
	Limits of Current ORDBMS	15
	Parallel Processing of UDF	17
	Two Step Parallel Aggregation of UDAF	17
	Partitioning Classes and Partitionable Functions	18
	Parallel Sorting as a Preprocessing Step for UDAF	21
	Extended Syntax for Function Registration	22
	Example Applications	24
	The UDAF MostöFrequent	24
	The UDAF RunningöAverage	25
	The UDAF Median	25
	Further Applications	26
	Plausibility Considerations Regarding Performance	28
Ch. 4	Intra-function Parallelism
	Compose/Decompose Operators for Intra-function Parallelism	34
	Compose/Decompose Operators	34
	Extensibility of Compose Operators by Combine Functions	36
	Application of Intra-function Parallelism	37
	Intra-function Parallelism for Function Pipelines	38
	Experimental Performance Study	39
	Experimental Scenario and Implementation	39
	Performance Results	41
Ch. 5	The Multi-operator Method
	Performance Problems with Complex UDF in Current ORDBMS	46
	The PBSM Algorithm as a Sophisticated UDP Implementation	47
	The Multi-operator Method as a New Technique to Implement Complex UDF	49
	The Multi-operator Method and Its Benefits	49
	A Multi-operator Implementation of the PBSM Algorithm	51
	Supporting the Multi-operator Method	53
	Executing Query Execution Plans	53
	Example for a Textual Specification of Query Execution Plans	55
	Parallel Evaluation	55
	Performance Evaluation	56
	Experimental Scenario	56
	Performance Results	62
Ch. 6	User-Defined Table Operators
	User-Defined Table Operators	68
	A Generalization Relationship for Row Types	68
	Defining and Implementing UDTO	69
	The Different Usages of the UDTO Concept	74
	Parallel Processing of Procedural UDTO	77
	Extension to Multiple Output Tables	80
	Example Applications for UDTO	81
	Computing a Spatial Join	81
	Different UDTO for the Same Predicate	85
	Computing the Median: An Aggregation Operator	89
	A UDTO for a Complex Aggregation	90
	Association Rule Mining	94
Ch. 7	Implementation of UDTO
	The MIDAS Prototype	106
	Architectural Overview	107
	Query Compilation and Execution	108
	The MIDAS System Tables	111
	UDSF in MIDAS	112
	Implementation of SQL Macros	113
	DDL Statements	113
	SQL Macro Expansion in DML Statements	115
	Expanding SQL Macros in Preprocessors and Middleware	116
	Implementation of Procedural UDTO	123
	Extensions to the SQL Compiler	123
	Extensions to the Optimizer and the Parallelizer	125
	Extensions to the Scheduler	126
	Extensions to the Execution Engine	126
	Extensions to Transaction Management	128
	Implementation of Input and Output Tables	131
	Optimization Issues for UDTO	134
	UDTO and Implied Predicates	134
	Estimating Costs and Selectivity of UDTO	135
	Application of Traditional Optimization Rules	137
	Using UDTO to Generate Alternative Execution Plans for UDF	138
	Evaluation of the Implementation	139
	Evaluation of SQL Macros	140
	Evaluation of Procedural UDTO	142
Ch. 8	Summary, Conclusions, and Future Work
	References	151
A.1	The Program sequentialöinvert	157
A.2	The Program parallelöinvert	158
A.3	The Query Execution Plan for the Spatial Join with SQL Macro	159