ITR: A Petabyte in Your Pocket

在2015年，每年花费几百美元，每个人都将拥有一个个人pb级的数据库（PetDB），可以从任何连接点，从高性能工作站到PDA的任何设备访问。每个人的PetDB将是他们对存在于任何地方的所有在线数字数据不断发展和定制的视图。它将存储和组织任何类型的数字数据，而不会丢失结构或信息。所有这些数据都可以按类型、内容、结构、关联以及多个分类和分组进行查询和排列。PetDB是我们称之为网络数据管理器（ndm）的新一代软件基础设施的一个例子。本研究的目的不是构建PetDB本身；而是设计和实现NDM技术，在此基础上可以很容易地构建petdb和其他应用程序。ndm与当前的数据库管理系统不同。它们关注于数据移动，而不是数据存储，并且必须处理任意类型的数据，而不必具有匹配的数据库模式。他们必须在数以万计的信息站点上执行查询，同时在快速变化的信息源上监视可能数以百万计的触发器。通过对XML查询、上下文文本索引和搜索、多触发器规划和数据流处理的初步工作，已经向ndm迈出了探索性的第一步。这项工作展示了ndm的一些基本功能。继续研究将解决必须解决的具有挑战性的基本问题，以扩展这些能力以利用ndm的全部能力。与此同时，通过分布式和并行实现，在用户数量、任务和站点数量以及数据量方面，将扩展ndm的运行规模。

In 2015, for a few hundred dollars a year, everyone will have a personal petabyte database (PetDB) that can be accessed from any point of connection, with any device from a high-powered workstation to a PDA. Each individual's PetDB will be their evolving and customized view of all the on-line digital data that exists anywhere. It will store and organize any kind of digital data, without losing structure or information. All this data will be queryable and arranged by type, content, structure, association and multiple categorizations and groupings. The PetDB is an example of what could be done with a new generation of software infrastructure we term Net Data Managers (NDMs). The object of this research is not to build a PetDB per se; but to design and implement the NDM technology upon which PetDBs and other applications could be readily built. NDMs are a departure from current database management systems. They focus on data movement, rather than data storage and must handle data of arbitrary types, without necessarily having a matching database schema. They must execute queries over tens of thousands of information sites, while monitoring possibly millions of triggers over rapidly changing information sources. The first exploratory steps have been taken towards NDMs with initial work on XML querying, text in context indexing and searching, multi-trigger planning and data stream processing. This work has demonstrated some of the basic capabilities of NDMs. Continued research will address the challenging fundamental problems that must be solved to extend these capabilities to exploit the full capabilities of NDMs. At the same time, the scale at which NDMs operate will be extended, in terms of numbers of users, tasks and sites, as well as data volumes, through distributed and parallel implementations.