Introduction / 1: |
Background / 1.1: |
Parallel Database Systems / 1.2: |
Computation Model / 1.2.1: |
Engineering Model / 1.2.2: |
About this Manuscript / 1.3: |
Bibliography |
Request Manager / I: |
Designing an Optimizer for Parallel Relational Systems / 2: |
Overall Design Issues / 2.1: |
Design a Simple Parallel Execution Model / 2.2.1: |
The Two-Phase Approach / 2.2.2: |
Parallelizing is Adding Information! / 2.2.3: |
Two-Phase versus Parallel Approaches / 2.2.4: |
Parallelization / 2.3: |
Kinds of Parallelism / 2.3.1: |
Specifying Parallel Execution / 2.3.2: |
Search Space / 2.4: |
Slicing Hash Join Trees / 2.4.1: |
Search Space Size / 2.4.2: |
Heuristics / 2.4.3: |
The Two-Phase Heuristics / 2.4.4: |
Cost Model / 2.5: |
Exceptions to the Principle of Optimality / 2.5.1: |
Resources / 2.5.2: |
Skew and Size Model / 2.5.3: |
The Cost Function / 2.5.4: |
Search Strategies / 2.6: |
Deterministic Search Strategies / 2.6.1: |
Randomized Strategies / 2.6.2: |
Conclusion / 2.7: |
New Approaches to Parallel Join Utilizing Page Connectivity Information / 3: |
The Environment and a Motivating Example / 3.1: |
The Methodology / 3.3: |
Definition of Parameters / 3.3.1: |
The Balancing Algorithm / 3.3.2: |
Schedules for Reading Join Components and Data Pages / 3.3.3: |
Performance Analysis / 3.4: |
The Evaluation Method / 3.4.1: |
Evaluation Results / 3.4.2: |
Concluding Remarks and Future Work / 3.5: |
A Performance Evaluation Tool for Parallel Database Systems / 4: |
Performance Evaluation Methods / 4.1: |
Analytical Modeling / 4.2.1: |
Benchmarks / 4.2.2: |
Observations / 4.2.3: |
The Software Testpilot / 4.3: |
The Experiment Specification / 4.3.1: |
The Performance Assessment Cycle / 4.3.2: |
The System Interface / 4.3.3: |
The Software Testpilot and Oracle/Ncube / 4.4: |
Database System Performance Assessment / 4.4.1: |
The Oracle/Ncube Interface / 4.4.2: |
Preliminary Results / 4.5: |
Load Placement in Distributed High-Performance Database Systems / 4.6: |
Investigated System / 5.1: |
System Architecture / 5.2.1: |
Load Scenarios / 5.2.2: |
Trace Analysis / 5.2.3: |
Load Setup / 5.2.4: |
Load Placement Strategies Investigated / 5.3: |
Scheduling Strategies for Transactions / 5.4: |
Simulation Results / 5.5: |
Influence of Scheduling / 5.5.1: |
Evaluation of the Load Placement Strategies / 5.5.2: |
Lessons Learned / 5.5.3: |
Decision Parameters Used / 5.5.4: |
Conclusion and Open Issues / 5.6: |
Parallel Machine Architecture / II: |
Modeling Recovery in Client-Server Database Systems / 6: |
Uniprocessor Recovery and Formal Approach to Modeling Recovery / 6.1: |
Basic Formal Concepts / 6.2.1: |
Logging Mechanisms / 6.2.2: |
Runtime Policies for Ensuring Correctness / 6.2.3: |
Data Structures Maintained for Efficient Recovery / 6.2.4: |
Restart Recovery--The ARIES Approach / 6.2.5: |
LSN Sequencing Techniques for Multinode Systems / 6.3: |
Recovery in Client-Server Database Systems / 6.4: |
Client-Server EXODUS (ESM-CS) / 6.4.1: |
Client-Server ARIES (ARIES/CSA) / 6.4.2: |
Shared Nothing Clients with Disks (CD) / 6.4.3: |
Summary of Recovery Approaches in Client-Server Architectures / 6.4.4: |
Parallel Strategies for a Petabyte Multimedia Database Computer / 6.5: |
Multimedia Data Warehouse, Databases, and Applications / 7.1: |
Three Waves of Multimedia Database Development / 7.2.1: |
National Medical Practice Knowledge Bank Application / 7.2.2: |
Massively Parallel Architecture, Infrastructure, and Technology / 7.3: |
Parallelism / 7.3.1: |
Teradata-MM Architecture, Framework, and New Concepts / 7.4: |
Teradata-MM Architecture / 7.4.1: |
Key New Concepts / 7.4.2: |
SQL3 / 7.4.3: |
Federated Coordinator / 7.4.4: |
Teradata Multimedia Object Server / 7.4.5: |
Parallel UDF Execution Analysis / 7.5: |
UDF Optimizations / 7.5.1: |
PRAGMA Facility / 7.5.2: |
UDF Value Persistence Facility / 7.5.3: |
Spatial Indices for Content-Based Querying / 7.5.4: |
The MEDUSA Project / 7.6: |
Indexing and Data Partitioning / 8.1: |
Standard Systems / 8.2.1: |
Grid Files / 8.2.2: |
Dynamic Load Balancing / 8.3: |
Data Access Frequency / 8.3.1: |
Data Distribution / 8.3.2: |
Query Partitioning / 8.3.3: |
The MEDUSA Architecture / 8.4: |
Software / 8.4.2: |
Grid File Implementation / 8.4.3: |
Load Balancing Strategy / 8.4.4: |
MEDUSA Performance Results / 8.5: |
Test Configuration / 8.5.1: |
Transaction Throughput / 8.5.2: |
Speedup / 8.5.3: |
Load Balancing Test Results / 8.5.4: |
Conclusions / 8.6: |
Partitioned Data Store / III: |
System Software of the Super Database Computer SDC-II / 9: |
Architectural Overview of the SDC-II / 9.1: |
Design and Organization of the SDC-II System Software / 9.3: |
Parallel Execution Model / 9.3.1: |
I/O Model and Buffer Management Strategy for Bulk Data Transfer / 9.3.2: |
Process Model and Efficient Flow Control Mechanism / 9.3.3: |
Structure of the System Software Components / 9.3.4: |
Evaluation of the SDC-II System / 9.4: |
Details of a Sample Query Processing / 9.4.1: |
Comparison with Commercial Systems / 9.4.2: |
Data Placement in Parallel Database Systems / 9.5: |
Overview of Data Placement Strategies / 10.1: |
Declustering and Redistribution / 10.2.1: |
Placement / 10.2.2: |
Effects of Data Placement / 10.3: |
STEADY and TPC-C / 10.3.1: |
Dependence on Number of Processing Elements / 10.3.2: |
Dependence on Database Size / 10.3.3: |
Contributors / 10.4: |
Introduction / 1: |
Background / 1.1: |
Parallel Database Systems / 1.2: |
Computation Model / 1.2.1: |
Engineering Model / 1.2.2: |
About this Manuscript / 1.3: |