Data Warehousing and Decision Support

在企业、政府和生活的各个方面，规划都很重要，因此需要能够处理“决策支持查询”的数据库管理系统。这些复杂的查询涉及到大量数据的聚合。通常，为了请求决策支持查询而创建特殊的“数据仓库”或来自各种来源的信息副本，而原始数据库则用于事务处理。一些新的数据库产品实现了“数据集”，这是一种特殊的数据组织，可以促进通用形式的决策支持查询。一个重要的新研究问题是数据立方体设计，对于给定的存储量，实现一组选定的视图来优化平均查询时间是可行的。为数据集选择物化视图是一个主要的研究挑战。更一般地说，当仓库中的视图随着底层数据库的变化而被维护时，希望在仓库中有一个视图选择，以便它们都可以被更新，而不必在底层数据库上发出查询（所谓的视图的“自维护性”）。自我维护的仓库提出了另一个研究挑战。最后，决策支持查询本身通常可以通过最近发现的技术进行优化，并且似乎需要研究一些新技术。

In business, government, and all aspects of life where planning is important, there is a need for database management systems that can handle `decision-support queries.` These complex queries involve the aggregation of very large amounts of data. Often, special `data warehouses,` or copies of information from a variety of sources, are created for the purpose of asking decision-support queries, while the original databases are used for transaction processing. Some new database products implement `data cubes,` special organizations of data that facilitate common forms of decision-support queries. One important new research problem is data-cube design, where it is feasible to materialize a selected set of views to optimize average query time, for a given amount of storage. Selection of materialized views for data cubes is a major research challenge. More generally, when views in a warehouse are maintained as the underlying databases change, it is desirable to have at the warehouse a selection of views so that they can all be updated without having to issue queries at the underlying databases (so-called `self-maintainability` of views). Self-maintainable warehouses present another research challenge. Finally, decision-support queries themselves can often be optimized by techniques recently discovered and, it appears, some new techniques that need investigation.