Collaborative Research: Affinity Directed Mobility for Location Independent Data Access

目前，人们使用各种分散的工具来访问和操纵信息。研究人员使用台式计算机作为我们的主要信息工具、服务器、服务器网络或ISP来提供文件系统支持、计算密集型计算和其他系统支持功能。当他们需要一种便携的信息工具时，他们使用笔记本电脑，这种电脑可能有也可能没有网络连接。当信息需求不大，移动需求最明显时，研究人员使用各种其他设备，如掌上电脑、个人数字助理、移动电话和寻呼机。在这种情况下，软件版本管理效率低下，许可也很复杂。确保在大量计算机上安装每个所需应用程序的相同（和最新）版本可能会消耗大量的系统支持人员资源，即使这些计算机可以从同一网络访问。保护信息不受未经授权的访问也同样困难。数据、应用程序和设备的这种碎片化的结果是信息工具的集合越来越复杂和难以管理，它们之间的通信效率低下。大量需求、大量通信基础设施和高性能、低功耗计算资源的融合要求研究人员探索更好的替代方案。Bifrost位置独立计算项目旨在提供一个灵活和全面的信息访问环境。Bifrost的功能是提供与位置和设备无关的数据访问。Bifrost中的数据既包括信息，也包括用于操作该信息的应用程序。Bifrost使用数据之间、用户和数据之间的亲缘关系来做出关于何时何地移动数据的适当决策。我们把这种方法称为亲和定向迁移。该项目的核心研究问题是移动性管理（我们如何移动数据和线程以支持用户和设备的移动性）、数据管理（我们如何表示、访问、更新和保护信息）和应用程序管理（我们如何提供跨各种设备的公共应用程序基础）。与之前的移动数据管理方法相比，Bifrost预计在3-5年内，网络连接将成为常态，即使对于高度移动的计算设备也是如此。在这种情况下，无论位置或计算设备如何，用户如何访问远程个人数据的问题至少与为可能的断开连接做准备一样重要。研究人员建议设计、实现、部署和评估Bifrost位置独立计算系统的两个校园原型。Bifrost的设计代表了信息访问和操作的新范式。我们建议将管理信息的方式与管理访问这些信息的应用程序的方式集成在一起。Bifrost的设计原则之一是访问数据的能力应该在所有平台上保持一致，包括便携式计算设备。这种方法与目前与大多数手持计算设备相关联的精简操作系统+精简应用程序+有限数据集计算模型形成鲜明对比。除了对直接参与该项目的学生的教育效益外，研究人员还希望该项目能以更广泛的方式为学术界做出贡献。他们将把这项研究纳入科罗拉多大学和康奈尔大学的操作系统和分布式系统课程中。该项目的所有软件和其他研究产品将通过万维网随时提供。

Information is currently accessed and manipulated using a variety of fragmented tools. The researchers use a desktop computer as our primary information tool, a server, network of servers, or an ISP to provide file system support, computationally intensive computing, and other system support functions. They use a notebook computer, which may or may not have network access, when they desire a portable information tool. When information needs are modest, and mobility needs most apparent, the researchers use a variety of other devices such as handheld computers, personal digital assistants, cellular telephones, and pagers. Software version management in this situation is inefficient, and licensing is complex. Ensuring that the same (and latest) version of each desired application program is installed on a large number of computers can consume a significant amount of system support staff resources, even if these computers are accessible from the same network. The protection of information from unauthorized access is similarly difficult.The result of this fragmentation of data, applications, and devices is an increasingly complex and unmanageable collection of information tools that communicate with each other ineffectively. The convergence in time of substantial need, substantial communication infrastructure, and high performance, low power computing resources challenges the researchers to explore a better alternative. The Bifrost location independent computing project seeks to provide a flexible and comprehensive information access environment. The function of Bifrost is to provide location and device independent access to data. Data in Bifrost encompasses both information and the applications used to manipulate that information. Bifrost uses affinity between data, and between users and data, to make appropriate decisions about when and where to move data. We refer to this approach as affinity directed mobility.The core research issues of the project are mobility management (how we move data and threads to support user and device mobility), data management (how we represent, access, update, and protect information), and application management (how we provide a common applications base across a variety of devices). In contrast to previous approaches to mobile data management, Bifrost anticipates that in 3-5 years network connectivity will be the norm, even for highly mobile computing devices. In this situation, the problem of how users can access remote personal data, regardless of location or computing device, becomes at least as important as planning for a possible disconnection.The researchers propose to design, implement, deploy and evaluate a two-campus prototype of the Bifrost location independent computing system. The Bifrost design represents a new paradigm for information access and manipulation. We propose to integrate the way in which information is managed with the way in which the applications that access this information are managed. One of the design principles of Bifrost is that the ability to access data should be consistent across all platforms, including portable computing devices. This approach is in stark contrast to the stripped down operating system plus stripped down applications plus limited data set computing model currently associated with most hand-held computing devices.In addition to the educational benefits to the students directly involved in the project, the researchers intend that this project contribute in broader ways to the academic community. They will incorporate this research into the operating systems and distributed systems classes taught at the University of Colorado and Cornell University. All software and other research products of this project will be made readily available via the world wide web.