Data Warehouse Performance
Data Warehouse Performance
复制标题
数据仓库性能
DOI：
10.4018/978-1-59140-557-3.ch061
复制
发表时间：
2009
期刊：
影响因子：
--
通讯作者：
Zu
中科院分区：
文献类型：
--
作者：
Beixin Lin;Yu Hong;Zu
数据仓库是一个大型电子信息存储库，由企业随着时间的推移以结构化的方式生成和更新，以帮助商业智能和支持决策。数据仓库中存储的数据是非易失性和时变性的，并以支持决策的方式按主题组织(Inmon等人，2001年)。数据仓库作为商业智能报告的骨干技术越来越多地被企业所采用，查询性能成为数据仓库成功实施的关键。根据Appfluent Technology对358家企业进行的一项关于报告和最终用户查询工具的调查，数据仓库的性能显著影响商业智能(BI)系统的投资回报(ROI)，并直接影响系统的底线(Appfluent Technology，2002)。即使在某些情况下，很难用ROI或美元数字来衡量BI项目的好处，但管理团队仍然渴望通过使用BI解决方案来获得“单一版本的真相”，为战略和战术决策提供更好的信息，并通过使用BI解决方案来更有效地执行业务流程(Eckerson，2003)。随着时间的推移，数据量急剧增加，数据质量参差不齐，可能会对数据仓库的性能产生不利影响。一些数据可能会随着时间的推移变得过时，并可能与仍然有效的决策数据混合在一起。此外，收集数据通常是为了满足潜在的要求，但可能永远不会使用。数据仓库还包含外部数据(例如，人口统计、心理特征等)。支持各种预测性数据挖掘活动。所有这些因素都促成了数据量的大规模增长。因此，即使是一个简单的查询也可能成为处理系统索引的负担并导致系统索引溢出(Inmon等人，1998)。因此，探索性能调优技术成为数据仓库管理中的一个重要课题。
A data warehouse is a large electronic repository of information that is generated and updated in a structured manner by an enterprise over time to aid business intelligence and to support decision making. Data stored in a data warehouse is non-volatile and time variant and is organized by subjects in a manner to support decision making (Inmon et al., 2001). Data warehousing has been increasingly adopted by enterprises as the backbone technology for business intelligence reporting and query performance has become the key to the successful implementation of data warehouses. According to a survey of 358 businesses on reporting and end-user query tools, conducted by Appfluent Technology, data warehouse performance significantly affects the Return on Investment (ROI) on Business Intelligence (BI) systems and directly impacts the bottom line of the systems (Appfluent Technology, 2002). Even though in some circumstances it is very difficult to measure the benefits of BI projects in terms of ROI or dollar figures, management teams are still eager to have a “single version of the truth,” better information for strategic and tactical decision making, and more efficient business processes by using BI solutions (Eckerson, 2003). Dramatic increases in data volumes over time and the mixed quality of data can adversely affect the performance of a data warehouse. Some data may become outdated over time and can be mixed with data that are still valid for decision making. In addition, data are often collected to meet potential requirements, but may never be used. Data warehouses also contain external data (e.g. demographic, psychographic, etc.) to support a variety of predictive data mining activities. All these factors contribute to the massive growth of data volume. As a result, even a simple query may become burdensome to process and cause overflowing system indices (Inmon et al., 1998). Thus, exploring the techniques of performance tuning becomes an important subject in data warehouse management.