Adaptive Synchronization Framework Supporting Device-independent Mobile Computing

无线网络和便携式信息设备的最新技术进步催生了移动计算的新范式，使携带便携式设备的用户能够访问和更新数据，而无论其物理位置或移动行为如何。然而，这项技术的使用仍将受到限制，除非我们开发出必要的概念、理论和基础设施，将移动性和断开连接无缝地集成到日常网络计算中。要让移动计算变得普及，还需要进行广泛的研究。目前的研究工作不支持移动数据访问和更新的充分自动化和管理。通常对应用程序、移动数据的来源或移动设备的特殊性进行假设。由于移动设备和移动数据之间的紧密耦合，缺乏无处不在的数据访问。因此，在没有花费大量精力复制和重新囤积移动设备的情况下，用户不能更换移动设备。此外，当前的解决方案不支持从诸如文件系统、数据库服务器、网络服务器等不同数据源的移动访问和囤积。当前的研究也局限于基本同步方案，其没有考虑移动数据的一致性和最新的可变和个体需求。最后，当前的移动事务模型在提交成功率方面效率不高，特别是在长时间断开连接的情况下。这项提议是基于对移动计算、操作系统和数据仓库的持续研究。总体目标是通过自动将来自多个异类数据源的各种数据存储到移动设备中，使更广泛的用户和应用程序能够使用移动计算，并促进移动设备和源所在的固定网络之间的复杂同步。具体目标是：开发和评估基于Coda文件系统的三层体系结构，该体系结构通过用于存储用户工作集的数据仓库提供移动数据和移动设备之间的独立性。开发算法，用于在移动设备上自动和增量地存储数据，并将移动设备的内容与仓库中的工作集和原始源同步。根据移动数据项的可编程和有条件的一致性规范，开发用于维护用户工作集的新同步技术。不同的数据项可能有不同的一致性要求，因此必须进行不同的同步。开发一个新的移动事务模型，该模型可以保证在断开连接期间执行的事务更新极有可能在重新连接时提交。我们的目标是在现有结果的基础上开发架构和算法，使移动环境中的智能囤积和同步成为现实。我们认为，实现这样一个框架是朝着使移动计算成为广大受众可行的做法迈出的重要和必要的一步；这一步有朝一日可能导致我们实现无处不在的计算。在我们的愿景中，应该允许用户切换到任何移动设备来连接到固定网络并随身携带必要的数据，而不必担心囤积、同步和其他低水平的负担。鉴于我们之前在移动计算、分布式系统和数据仓库领域的经验和研究成果，我们相信可以成功地应用数据仓库技术来管理重要的用户数据，并支持根据频繁的更新对这些数据进行增量维护。

Recent technological advances in wireless networks and portable information appliances have engendered the new paradigm of mobile computing, enabling users carrying portable devices to access and update data regardless of their physical location or movement behavior. However, the use of this technology will remain limited unless we develop the necessary concepts, theory and infrastructure that can seamlessly integrate mobility and disconnection into everyday networked computing. Extensive research is needed in order for mobile computing to become pervasive. Current research efforts do not support sufficient automation and management of mobile data access and update. Assumptions are usually made regarding the application, the source of mobile data, or the particularities of the mobile device. Ubiquitous data access is lacking due to the tight coupling between the mobile device and the mobile data. Users are therefore not allowed to switch mobile devices without spending a lot of effort on copying and re-hoarding. Additionally, current solutions do not support mobile access and hoarding from heterogeneous sources of data such as file systems, database servers, web servers, etc. Current research is also limited to basic synchronization schemes that do not take into consideration the variable and individual needs of consistency and up-to-dateness of mobile data. Finally, current mobile transaction models are not efficient in terms of successful commit rate, especially under prolonged periods of disconnection. This proposal is based on ongoing research on mobile computing, operating systems and data warehousing. The overall goal is to make mobile computing available to a broader range of users and applications by automating the hoarding of a wide variety of data from multiple heterogeneous sources into mobile devices, and to facilitate sophisticated synchronization between the mobile devices and the fixed networks where the sources reside. The specific goals are:Develop and evaluate a three-tiered architecture based on the Coda file system that provides independence between the mobile data and the mobile devices via a data warehouse for storing the user's working set. Develop algorithms for automatically and incrementally hoarding data on mobile devices and for synchronizing the contents of the mobile device with the working set in the warehouse and the original sources.Develop new synchronization techniques for maintaining the user's working set, based on programmable and conditional consistency specifications of mobile data items. Different data items may have different consistency requirements, and must therefore be synchronized differently.Develop a new model of mobile transactions that can guarantee that transactional updates performed during disconnection are highly likely to be committed upon reconnection.Our goal is to build upon existing results to develop the architecture and algorithms to make smart hoarding and synchronization in mobile environments a reality. We see the realization of such a framework as an important and necessary step towards making mobile computing a viable practice for a broad audience; a step that some day may lead us to the realization of ubiquitous computing. In our vision, users should be allowed to switch to any mobile device to connect to the fixed network and carry the necessary data with them, without having to worry about hoarding, synchronization, and other low-level burdens. Given our prior experience and research results in the areas of mobile computing, distributed systems, and data warehousing, we believe that we can successfully apply data warehousing technology to manage important user data as well as support the incremental maintenance of this data in light of frequent updates.