RIA: Scalable Technologies for Highly Parallel Database Management System Implementation

RIA：用于高度并行数据库管理系统实现的可扩展技术联邦HPCC计划要求开发可扩展的并行计算系统，能够在非常大的数据集上维持每秒数万亿次的操作。为了应对这一挑战，数据管理系统所需的可扩展并行处理技术的分析和实验研究已被确定为许多重大挑战研究的关键需求。本研究旨在研究这些问题，特别是研究可扩展的存储结构和高效的查询处理技术。该工作包括理论分析和系统实现。该项目的预期结果包括：(1)多维分布式文件结构，为处理非常大的数据集提供了高效的环境；(2)多维数据聚类环境下具有动态负载均衡能力的高效连接策略；(3)一种考虑负载平衡成本的有效查询优化技术。这项研究的产品包括一个高级查询优化器，它既可以作为可行性演示的工具，也可以作为未来数据库机器开发的平台。这项研究的意义在于推进了实现大型并行数据库系统所需的可扩展软件技术。其结果将对为重大挑战问题提供有效的解决方案产生影响。

RIA: Scalable Technologies for Highly Parallel Database Management System Implementation The Federal HPCC Initiative calls for the development of scalable parallel computing systems capable of sustaining trillions of operations per second on very large datasets. In meeting this challenge, analytic and experimental investigations of scalable parallel processing techniques required for data management systems have been identified as critical needs for many grand challenge studies. This research is aimed at studying these issues, in particular, to investigate scalable storage structures and efficient query processing techniques. The work involves both theoretical analysis and system implementation. The expected results of this project include: (1) a multidimensional distributed file structure that facilitates an efficient environment for handling very large data sets; (2) an efficient join strategy with dynamic load balancing capabilities for the multidimensional data declustering environment; and (3) an effective query optimization technique that takes the costs of load balancing into consideration. Products of this investigation include an advanced query optimizer that serves both as a vehicle for feasibility demonstrations, and a platform for future database machine development. The significance of this research is in advancing the scalable software technologies required for the implementation of large parallel database systems. The results will have an impact on providing efficient solutions to the grand challenge problems.