Research Initiation: Relational Databases in a Hierarchical Design Environment

作为管理日益复杂的工程设计的一种方式，设计自动化社区正在转向使用分层设计工具：这些工具对正在设计的复杂结构的多层表示进行操作，以便在用户明确请求查看设计的较低、更详细的方面之前看不到这些方面。这项研究将调查分层、高度交互的设计环境对存储设计信息的数据库管理系统的性能要求的影响。其他人完成的工作表明，扩展关系数据库在具有(1)大量使用主存缓存、(2)非关系实现和(3)分层设计工具的设计环境中可以表现出令人满意的数据检索性能。本研究的目标是对分层设计工具的数据检索和更新性能需求进行更细粒度的检查，并测试使用扩展的关系数据库是否可以获得令人满意的更新性能。为此，首席调查员将为层次化VLSI布局编辑Magic提供工具；收集专家设计师使用Magic的协议；并对照适当的数据库访问成本模型分析这些协议。该项目将包括对一名本科生的支持。

As a way of managing the growing complexity of engineering designs, the design automation community is moving towards the use of hierarchical design tools: tools that operate on multi- layer representations of the complex structures being designed, so that lower, more detailed aspects of a design are not visible until the user specifically requests to see those aspects. This research will investigate the impact of a hierarchical, highly interactive design environment on the performance requirements for a database management system storing design information. Work completed by others shows that extended relational databases can exhibit satisfactory data retrieval performance in a design environment with (1) heavy use of main-memory caching, (2) a non-relational implementation, and (3) hierarchical design tools. The goal of this research is to develop a finer-grained examination of the data retrieval and update performance requirements of hierarchical design tools, and to test whether satisfactory update performance can be obtained using an extended relational database. To this end, the principal investigator will instrument Magic, a hierarchical VLSI layout editor; collect protocols of Magic's use by expert designers; and analyze these protocols against appropriate models of database access costs. This project will include support for one undergraduate student.