SGER: Demonstrating the Feasibility of Semantic Reconfiguration and Semantic Checkpointing

该项目解决了在执行过程中被迫或可能受益于使用替代资源的应用程序的系统支持。 其目的是在应用程序重组过程中，包括应用程序的重大重新设计过程中，为应用程序用户提供尽可能多的连续性。 这个项目是一个SGER，调查并提供了两个关键问题的概念证明：语义重构和语义检查点。 语义重配置是基于备用资源重新组织和重新实现应用程序的过程，同时保留应用程序的整体语义。 语义检查点是在应用程序本身的语义级别捕获和恢复状态的过程，与捕获底层硬件或操作系统的状态相反。 为此，该项目基于具有特定一致性要求的分布式数据库应用程序，在低容量和高容量服务器和网络的组合中移动。最初的手工系统包括：应用程序的规范和检查点的状态，组件组合，检查点状态的转换，以及检查点本身。 这项工作的贡献是：1）设计和构建关键算法来实现语义重构和语义检查点，2）理解这两个任务的泛化和自动化路径，以及3）识别和构建完成自动化转换所需的其他关键组件。 随着移动的应用变得越来越普遍，这种系统化的方法将变得越来越重要。

This project addresses the provision of system support for applications that are forced or may benefit from use of alternate resources during their execution. The objective is to provide the application user as much of a sense of continuity as possible across reorganization of the application, including significant remodularization of the application. This project is an SGER that investigates and provides a proof of concept of two key problems: semantic reconfiguration and semantic checkpointing. Semantic reconfiguration is the process of reorganizing and remodularizing an application, based on alternate resources, while retaining the overall semantics of the application. Semantic checkpointing is the process of capturing and restoring state at the semantic level of the application itself, in contrast with capturing the state of the underlying hardware or operating system. To this end, the project is based on a distributed database application with specific consistency requirements, moving among combinations of low and high capacity servers and networks. The initial, handcrafted system includes: specification of the application and state to be checkpointed, component composition, conversion of checkpointing state, and the checkpointing itself. The contributions of this work are to 1) design and build the key algorithms to achieve semantic reconfiguration and semantic checkpointing, 2) understanding a path toward generalization and automation of these two tasks, and 3) identify and frame other key components required for complete automated transitions. As mobile applications become more prevalent, such a systematic approach will become increasing critical.