Most organisations need to combine data from various sources to inform their decisions. Nonetheless, the process of analysing data against transactional data can lead to the sabotage of a business system, and for this reason, businesses generally put data in different repositories for use in their analytical work (Inmon, 2015). Two terms given to these storehouses are data warehouse and mart.
A data warehouse is a federated repository for data accumulated from different sources of the business system. It is created by capturing data from different sources that are combined and made available for access and analysis (Kimball & Ross, 2012). Data warehousing has two approaches, namely, bottom-up and top-down. The bottom-up approach starts with building data marts then unites them into a single warehouse. The top-down approach spins off data marts for particular groups after the creation of the data warehouse.
A data mart is a repository of data designed for use by a particular group of workers in an organisation. Both terms can sometimes be used in the same context, but the difference is that the data warehouse holds an organisation’s full information combined from different sources while the data mart holds information meant for use by a specific group of workers in an organisation. The data mart of business has the subgroups of the data warehouse (Geary et al., 2017).
A data warehouse holds a collection of the organisation’s historical data separated from its operations. The data is used when decisions need to be made and is assembled after collection from different sources; this ensures that such data is accurate and that there is correct timing of information. The data in a warehouse entails an accumulation of data collected from different functional units (Kimball, 2014). Before the data in a warehouse is stored, it is cleaned, checked and eventually integrated into a particular unit. The warehouse is managed by a central unit in the business. On the contrary, a data mart forms a unit of the warehouse. It is specifically subject oriented and meets specific needs such as human resource management and finance.
One department controls data marts after assembling from different sources. Thus the major difference between both repositories is in their data sources and scope (Inmon, 2015). A data mart is usually about 100 gigabytes while a data warehouse is bigger. A data warehouse is more complex to design compared to a data mart.
Operational data is extracted for business intelligence for control and security and because the operational server performance is degraded by business intelligence analysis (Kimball, 2014). Operational data is designed to enable fast and reliable transactions. A data warehouse extracts data from different internal and external executive storages. After the data has been collected from different sources, it is cleaned at taken to the database management system where the metadata and the database warehouse meet; this is where the business intelligence tools get involved into the repository.
Operational data does not support business intelligence because it reduces the system’s performance which requires huge processing; a business data warehouse may take time to be fully formed (Moody & Kortink, 2013). For example, a 2008 report by the New York Times showed that Wal-Mart created several hundred data marts but had to hire an IT expert to combine the different systems into an integrated data warehouse.
Geary, N., Jarvis, B., Mew, C., & Gore, H. (2017). U.S. Patent No. 9,684,703. Washington, DC: U.S. Patent and Trademark Office.
Golfarelli, M., & Rizzi, S. (2017). Data warehouse design: Modern principles and methodologies. McGraw-Hill, Inc.
Inmon, W. H. (2015). Building the data warehouse. John Wiley & sons.
Kimball, R., & Ross, M. (2012). The data warehouse toolkit: the complete guide to dimensional modeling. John Wiley & Sons.
Kimball, R. (2014). The data warehouse lifecycle toolkit: expert methods for designing, developing, and deploying data warehouses. John Wiley & Sons.
Moody, D. L., & Kortink, M. A. (2013, June). From enterprise models to dimensional models: a methodology for data warehouse and data mart design. In DMDW (p. 5).
Ratings
