Diplomat: A System for Building and Maintaining Heterogeneous Database Alliances

许多大公司发现自己存在大量遗留数据库问题。它们的数据分布在许多语义相关的数据库中，这些数据库在物理上和逻辑上都是不相连的。到目前为止，有两种主要的互连异构数据库系统的方法：(a)点解决方案，其中支持数据库之间集中互连的软件系统是根据需要和特别的基础构建的；(b)全局解决方案，其中创建了一个包含大量数据库的公共逻辑数据库。（后者的一种变体是联邦方法，它涉及一个控制组件数据库之间所有交互的全局联邦管理器。）Diplomat项目选择了这两个极端之间的中间地带，并提供了支持称为“联盟”的多个松散数据库互连的技术。该框架还提供了在单个数据库之间和跨联盟之间加强一致性的机制。该项目借鉴了关于视图规范、数据库间约束和信息流的现有研究。这项研究证明了一个概念，即大型业务应用程序（如银行、保险和电信）的维护人员可以以一种经济、可发展的方式集成多个异构遗留数据库系统。

Many large corporations find themselves with a massive legacy database problem. Their data is distributed across many semantically related databases that are physically and logically disjoint. Until now, there have been two primary approaches for interconnecting heterogeneous database systems: (a) point solutions, in which software systems supporting focused interconnections between databases are constructed on an as-needed and ad hoc basis; and (b) global solutions, in which a common, logical database is created that encompasses a very large number of databases. (A variation of the latter is the federated approach, which involves a global federation manager that controls all interaction between the component databases.) The Diplomat project chooses a middle ground between these two extremes, and provides technology to support multiple loose database interconnections called `alliances`. The framework also provides mechanisms to enforce consistency between individual databases, and across alliances. The project draws on existing research on view specification, inter-database constraints, and information flow. This research serves as a proof of concept that maintainers of large business applications (such as banking, insurance, and telecommunications) can integrate multiple, heterogeneous, legacy database systems in a cost-effective, evolvable fashion.