大数据管理系统评测基准的挑战与研究进展

Challenges and Progress of Big Data Management System Benchmarks

数据库评测基准在数据库发展历史中的作用不可替代,而大数据环境中传统评测基准不敷应用。因此,从评测基准3要素,即数据、负载、度量体系入手,研究具有高仿真性、可适配性、可测量性的大数据管理系统评测基准,对大数据管理系统的研发和应用系统选型至关重要。基于此,在简要分析评测基准的基本要素和大数据管理系统发展过程的基础上,重点分析大数据管理系统的基准评测需求与挑战,然后通过社交媒体分析型查询评测基准BSMA,探讨了面向应用的大数据管理系统基准评测的设计和实现问题。

Database benchmarking has stimulated the development of data management systems and technologies. In big data environments, benchmarking should be revisited. Therefore, research on benchmarks for big data management systems is a key problem for big data research and applications. Benchmark design can be achieved from three different perspectives, i.e. data, workload, and performance measurements. After the brief introduction to these three elements and the progress of big data management system research, the requirements and challenges to benchmarking big data management systems were analyzed. Through the introduction to a benchmark for analytical queries over social media data, named as BSMA, the issues of design and implementation of a benchmark for big data management systems were discussed.