Examining Data Warehouse Use Cases

The first standard data warehouse use case is the data mart. Data marts are normally characterized by

two things: (1) their size (2) their focus. Data marts tend to be smaller in data volume than data

warehouses and they tend to be more narrow in their scope, oftentimes only holding data for a particular

area of a company or even a subset of an area in a company. A data mart’s update frequency (how often

its data is refreshed by daily/hourly transactional activity) depends on the need of the business area using

its information to make decisions. If key decisions necessitate that the most up-to-date data as possible be

present, real-time feeds into a data mart may be observed. Otherwise, daily or weekly refreshes will be

normative.

The real-time data warehouse has seen increased popularity in recent years, mainly due to the increased

desire to have the most current information as possible at the fingertips to navigate and outsmart the

competition. The real-time data warehouse’s attributes include constant resource contention between

incoming data refreshes and queries being made against the same set of data objects, hourly and daily

increases in storage, purge rituals of unneeded data, and an audience that can be either narrow or broad in

focus depending on the subject areas that reside in the warehouse.

The traditional data warehouse, as its name implies, is the use case most think about when it comes to

data warehousing. Usually sporting very large data volumes, having infrequent (defined as not hourly and

sometimes not daily) refresh rates, and serving a wide and varied audience, the traditional data warehouse

is what most businesses start with (after a data mart) when implementing a data store to use for business

intelligence purposes.

The historical data warehouse is somewhat new in nature and has been born out of semi-recent

mandates that require many businesses to keep large amounts of historical information at the ready for

government or other business-compliance purposes. The historical data warehouse typically has data

volumes that are multiples of traditional data warehouses, see semi-frequent data refreshes, but usually

have less query traffic than data marts, real-time warehouses, or traditional data warehouses.

The analytic OLTP warehouse application is what some see as a dangerous return to the days when

databases that were used for rapid transactional activity became the target for resource-intensive analytic

queries also. The mix of these workloads usually spelled death for applications needing rapid response

times to serve customers who demanded quick completions for their requests. The analytic OLTP

warehouse is usually a database that back-ends a standard OLTP application, but also contains objects

that are fed from the transactional objects and are designed to support business intelligence queries.

There may be other niche data warehouse installations, but the above represent the vast majority of what

are found in IT organizations.