CAJ | 학술논문

由于广域测量系统(wide area measurement system， WAMS)海量实时数据的大规模、高负荷并发访问，其采用的同步机制在很大程度上约束了系统的效率。针对于目前普遍使用的整体锁定机制带来的由于访问串行化导致的效率低下问题，提出了粒度可控并发访问同步算法(controllable granularity concurrency synchronization，CGCS)，该方法使用控制标志位建立基于子集超集依赖的锁定条件和先进先出等待队列，并可控锁定级别，实现对实时数据访问的互斥粒度粗细的任意控制，同时作用于整体文件、表、元组，进而任务可以最大化地并发执行。通过实验，给出了系统的并发处理能力和IO响应能力的测试过程和结果，证明了CGCS算法在并发高、访问散的情况下能充分发挥CPU的并行处理能力，使WAMS系统的效率得到大幅提升。
유우엄역측량계통(wide area measurement system， WAMS)해량실시수거적대규모、고부하병발방문，기채용적동보궤제재흔대정도상약속료계통적효솔。침대우목전보편사용적정체쇄정궤제대래적유우방문천행화도치적효솔저하문제，제출료립도가공병발방문동보산법(controllable granularity concurrency synchronization，CGCS)，해방법사용공제표지위건립기우자집초집의뢰적쇄정조건화선진선출등대대렬，병가공쇄정급별，실현대실시수거방문적호척립도조세적임의공제，동시작용우정체문건、표、원조，진이임무가이최대화지병발집행。통과실험，급출료계통적병발처리능력화IO향응능력적측시과정화결과，증명료CGCS산법재병발고、방문산적정황하능충분발휘CPU적병행처리능력，사WAMS계통적효솔득도대폭제승。
Due to large-scale and high-load concurrent access to massive real-time data of wide area measurement system (WAMS), the synchronization mechanism of WAMS constrains the system efficiency to great extent. For the problem of low efficiency caused by serialized access in common method of whole-locking mechanism, a controllable granularity concurrency synchronization (CGCS) algorithm was proposed in this paper. It is an approach of synchronization for in-memory data concurrent access using control flag to establish locking condition based on subset-superset dependence and first-in-first-out waiting queue which has controllable locking level. It can control mutex granularity of in-memory data access at any level, and enable it act on whole file, table, tuple level, and then the task can be executed concurrently to the most degree. By experiments, the process and results of test about system capability of concurrent access and IO response was given. The results show that CGCS algorithm gives full play to CPU ability to parallel process and largely improves the efficiency of WAMS in the high concurrency and dissipated access situation.