摘 要：现有的分布式实时协作系统多采用操作转换的方法来提供并发控制服务,但是在系统数据量很大时,系统性能不高.为解决这一问题,提出了一种新的并发控制方法madOPT:该方法利用对象所定义的操作的语义进行冲突解析,集数据对象的动态载入、数据传输和操作转换功能为一体.madOPT改进了dOPT(distributed operation transformation)算法在操作转换过程中对操作日志的遍历以提高并发控制效率,并将对象属性作为并发控制的粒度,使之支持对图形、图像对象的并发操作;同时结合组播代理,提高数据传输率,使系统整体性能得到改善.

Existing distributed real-time cooperative systems mainly use operation transformation technique to provide concurrency control service. But the performance of a run-time system will be degraded when the system produces a large amount of collaborative data. A novel concurrency control framework named madOPT based on collaborative data objects is then proposed in this paper. madOPT resolves conflicts by utilizing operation semantics of the accessed data objects. In addition, it integrates a data transmission framework with an operation transformation method and allows the running system to load the data objects dynamically. The enhancement of madOPT over dOPT (distributed operation transformation) is that it reduces the number of queries in operation log during an operation transformation. It takes object attributes as the granularity of concurrency control and thus can support graph and image objects. It also adopts multicast agent to improve the efficiency of data distribution as well as the performance of the system.

Supported by the National Natural Science Foundation of China under Grant No.60073011 (国家自然科学基金); the National High-Tech Research and Development Plan of China under Grant No.2001AA113150 (国家高技术研究发展计划(863)); the National High-Tech Research of Beijing Muni