用于透明计算的基于块粒度数据一致性方法

Block-based data consistency method for transparent computing

为了解决透明计算系统中的数据不一致问题,提出一种基于块粒度的数据一致性方法。该方法按需为每个用户创建独立的私有虚拟映像,将用户对系统虚拟映像的修改透明地转换为对用户私有映像的修改,并记录系统虚拟映像中被修改的数据和用户私有映像中相应数据的对应关系,保证每个用户能够访问各自最新的数据,从而解决了上述数据不一致问题。性能测试表明,采用该方法的透明计算系统性能优于同等配置的PC和其他类似系统,可以满足透明计算的需求。

A block-based data consistency method （BDCM） was developed to resolve data inconsistencies in transparent computing systems. The method creates PVDIs （private VDisk image） for users on demand and transparently redirects SVDI （System VDisk Image） update requests to the users＇ own PVDI. The system records the relationships between the updated data in the PVDIs and their corresponding version in the SVDIs to guarantee that the most recent version is correctly accessed by each user. The system effectively avoids data consistency violations. Performance measurements demonstrate that the transparent computing systems using this method are better than normal PC or similar systems with the same hardware configurations.