Research on Parallel Database Systems based on to Furctional Computation Model

为了支持数据库应用程序，如知识库和工程应用程序，有必要允许数据库设计人员指定新的基本操作和数据类型，并将它们集成到系统中。关键问题是为数据库设计人员提供实施特定操作和数据类型的设施，并使其能够整合到系统中。此外，这些操作应该高效地执行。为了支持高级数据库应用，数据库系统应该具有高效执行任意数据库操作的设施。研究了一种面向流的数据库基本操作并行处理方案(简称面向流方案)。该方案基于需求驱动的评估，结合流处理。我们为实现面向流的方案设计了几个基本操作。通过使用这些操作，各种数据库应用程序的数据库处理可以并行执行。在本研究中，我们还提出了面向流的并行处理方案的实现方法。在这种方法中，处理器之间的通信是通过消息传递机制来执行的。此外，我们给出了几个实验结果，并讨论了我们的实现方法的效率。

For supporting database applications, such as knowledge bases and engineering applications, it is necessary to allow the database designers to specify new basic operations and data types, and to integrate them into the system. The key issue is to provide database designers with facilities to implement specific operations and data types and to enable them to be integrated into the system. Furthermore, those operations should be executed efficiently.For supporting advanced database applications, database systems should have facilities to execute arbitrary database operations efficiently. We have studied a stream-oriented parallel processing scheme (stream-oriented scheme) for basic operations of databases. This scheme is based on the demand-driven evaluation incorporating stream processing. We have designed several basic operations for implementing the stream-oriented scheme. By using these operations, database processing for a wide variety of database applications can be performed in parallel.In this study, we have also presented an implementation method of the stream-oriented parallel processing scheme. In this method, the communication between processors is performed by using the message passing mechanism. Furthermore, we have shown several experimental results and discuss the efficiency of our implementation method.