Query Optimization in Object-Oriented Database Management Systems with Encapsulated Behavior

具有行为封装的面向对象数据库系统支持强大的数据抽象，但通常速度较慢。传统的数据库系统提供高效的查询处理，但数据建模功能有限。该研究项目寻求将抽象数据类型的能力与数据库优化技术相结合的新技术，使其成为语义强大且高效的查询和命令处理系统。如果对行为进行了封装，那么中央查询优化器就无法对整个查询的语义、行为和成本进行推理。对封装类型执行查询优化的关键是向可信的系统组件披露此类信息，并确保在披露点和评估点之间行为不会发生变化。由于类型和行为定义的语法分析可能只产生有限的性能改进，类型实现者将能够揭示其他相关信息。该项目包括一个面向对象的数据语言的定义，一个允许优化的代数查询语言，一个揭示机制，一个可扩展的优化器，以及一个验证所研究的概念和想法的软件原型。该研究的目标是开发一种关键技术，使具有封装行为的面向对象数据库系统能够获得与可扩展关系系统相当的性能，从而使此类系统成为管理大数据量的可行替代方案。

Object-oriented database systems with behavioral encapsulation support powerful data abstractions but are generally slow. Conventional database systems provide efficient query processing, but limited data modeling facilities. This research project seeks new techniques that combine the power of abstract data types with database optimization technology into a semantically powerful and efficient query and command processing system. If behavior is encapsulated, a central query optimizer cannot reason about semantics, behavior, and cost over an entire query. The key to performing optimization of queries over encapsulated types is in revealing such information to a trusted system component, and in ensuring that the behavior will not change between the points of revelation and evaluation. Since syntactic analysis of type and behavior definition may yield only limited performance improvement, the type implementor will be able to reveal other relevant information. The project includes the definition of an object-oriented data language, an algebraic query language that permits optimization, a revealing mechanism, an extensible optimizer, and a software prototype to validate the investigated concepts and ideas. The goal of the research is to develop a key technique that will allow object-oriented database systems with encapsulated behavior to attain performance comparable to extensible relational systems, thus making such systems viable alternatives for managing large data volumes.