随着互联网向传统行业渗透，以及传统领域的互联网化，后台事务处理系统面临着前所未有的性能和可扩展性压力。可扩展数据管理系统是解决新兴互联网级应用事务处理需求问题的主要发展方向，但由于日志记录本身的特点，现有技术不足以满足系统横向扩展后对日志记录高效、快速、高通量，以及事务恢复可靠、可用、即时的要求。项目针对具有高性能、高通量、高可靠、高可用、高一致性的事务处理需求，基于可扩展DBMS架构，系统研究高速日志记录和即时恢复的理论和方法；重点研究单节点日志记录优化、多节点日志同步策略、新型非易失内存的利用等关键技术；形成开源系统原型，通过基准评测，验证研究成果的效果；解决日志机制这一事务处理性能瓶颈问题；为落实我国关键信息系统“自主可靠、安全可控”发展战略提供技术支撑。申请人和研究团队在高速数据流处理、分布式数据管理技术、新型数据管理系统等方面有长期丰富的研究积累，具有顺利开展项目研究的良好基础。

The convergence of traditional domains and internet industries results in great pressure on backend transaction processing systems in the perspective of performance and scalability. Scalable data management systems are major players that may solve the problem. However, due to the requirements of consistency to database logs, existing logging mechanism cannot scale along with the system, thus become the bottleneck of the system. This project proposes to study fast logging and instant recovery theories and technologies. The research will focus on technologies of optimization of single-node logging, various log synchronization strategies, logging on non-volatile memory, etc. The technologies are to be integrated in open-source systems and evaluated by open benchmarks. The research aims at solving the bottleneck problem of logging, and supporting to build open and secure information systems. The research team has strong research background in streaming data processing, distributed data management, and database management systems with new architecture. The research plan is innovative and practical.