基于NoSQL集群数据库的下一代托卡马克实验数据存储技术研究

随着托卡马克实验向长脉冲发展和高速、高精度的数据采集的应用，托卡马克实验产生的数据量也在高速增长，ITER、CFETR等未来托卡马克装置产生的数据量将是目前装置的上百倍，且会随着装置的发展和持续运行不断增长。这对目前聚变领域现有的存储技术在高性能、高扩展性、高可靠性等方面提出了严峻的挑战，因此非常有必要对托卡马克实验数据存储技术进行前瞻性探索研究。本项目将探索利用新兴的NoSQL集群数据库技术，研究并实现一种容量与性能可方便地横向扩展，具有高并发性能、高可靠性的下一代托卡马克实验数据存储技术。所研究的数据存储技术不依赖任何特定的存储硬件，可大大降低实现成本。本项目有望给出基于NoSQL集群数据库的下一代托卡马克实验数据存储技术方案，为ITER、CFETR等未来托卡马克数据存储系统的研制提供重要参考。

As the tokamak discharge goes towards long pulse and faster digitizers are used in more sophisticated diagnostics, the scientific data generated during the experiment pulse is getting extremely larger. The future tokamaks like ITER and CFETR will produce hundreds times of data than today’s major tokamaks. Through their life, the demands will keep growing. Current available technologies in fusion community are hard to scale and cannot handle big data efficiently, they may barely meet the requirement of future tokamaks. This project attempts to find a solution for large tokamak scientific data archiving by using NoSQL cluster database technology. It will provide a storage and management system that can out performance the current available solutions and easily scale-out its capability, performance at low cost, but relies on no specific hardware technology. This technology will satisfy the continuously growing need for large scientific data storage of the future tokamaks and provide a good reference for future tokamak scientific data storage development.