智慧城市是未来城市治理的有效途径。交通数据处理和分析平台是智慧城市的重要基础设施，投资、开发、维护等成本大。随着交通数据规模的不断增加，传统的解决方案已经无法满足实际的应用需求，交通领域开始尝试使用诞生于互联网领域的大数据系统解决方案。 . 但是，交通应用计算需求的复杂性和大数据系统解决方案的多样性使平台的构建面临重重困难。当前，实施者在进行大数据系统选型时通常依赖于过往的经验或简单测试的结果。这种缺乏科学依据的决策往往会导致投资成本浪费，开发不顺利，运行维护困难等严重的后果。 . 针对交通大数据系统解决方案构建过程中遇到的问题，本课题尝试从大数据系统基准测试，性能评估和负载分析的角度开展研究，并构建一套针对交通大数据系统的基准测试工具和一套负载分析工具。研究成果能够准确综合评测不同解决方案的性能，能耗，可扩展性等多个指标，指导构建高效能的交通大数据系统解决方案。

Traffic data analysis platform is an important infrastructure for the smart city. How to build, develop and maintain this platform needs a huge amount of cost. As the data produced in traffic transportation systems explosive growth，the traditional IT solutions can not satisfy the realistic requirements. Thus, the operators in traffic field try to adopt the new emerging big data system solutions born in Internet area.. However, it is difficult to select an appropriate solution and build a platform because of the diverse computation requirements and big data system solutions. Currently, the operators selecting solutions depend on their own experiences or simple evaluation results. They make a decision without a reasonable foundation, leading to the waste of investment, unsmooth development and boring maintenance.. This work is trying to address this issue from benchmarking, performance evaluation and workload characterization for big data systems. We expect to design a benchmark suite and a workload characterization tool. The benchmark suite is able to comprehensively comparison different solutions in terms of performance, cost/performance, development efficiency, scalability, and maintenance cost. The workload characterization tool is used to analyze the program behavior of big data application and find the potential performance bottlenecks. These tools can help the operator to build an appropriate big data solution for traffic data analysis platform, obviously saving the cost of investment, development and maintenance