Data and its analysis form the foundation for sound business decisions. The data processing method known as Online Analytical Processing (OLAP) encompasses technologies, methods, and tools that enable the immediate analysis of multidimensional information. OLAP is classified as an analytical information system and falls under hypothesis-driven analytical methods. With the help of OLAP, an analysis can be conducted from various perspectives, with a focus on the multidimensional view of data.
The way OLAP works is best illustrated by an example: An online store sells ten different types of tea. The relevant data primarily includes the sales revenue for each type over a specific period, such as per month. The types of tea are further divided into two product groups—herbal teas and black teas—which in turn generates information on how much revenue each product group generates. The time factor is also hierarchically subdivided into years, quarters, and months. The teas are sold in the DACH region—that is, in Germany, Austria, and Switzerland. This data is also included, revealing where each tea generates how much revenue—location-based data can be further hierarchically organized by incorporating the individual regions of these countries into the existing data. In addition, there are further options for extracting data sets in detail: For example, individual orders can be broken down by order type—whether they were placed by phone, via an order form, or directly in the online store.
Important: Even though the data sets may seem simple at first glance, multidimensional database analysis is quite challenging. Such an analysis can be carried out as part of a pattern analysis—known as data mining—or via a data mart, which is a partial database analysis within a data warehouse. Typically, such a data processing system retrieves the required data from a data warehouse—in part because the information usually originates from various operational database systems within a company.
OLAP in Intralogistics
When logistics need to be optimized, an analysis is conducted to determine where it would be most cost-effective to store the tea varieties or which variety must be kept in stock at which location and in what quantity. The necessary key performance indicators are analyzed from various perspectives, which are also referred to as dimensions. Along with the hierarchies described above, they represent another typical concept.
If, for example, you want to know how much herbal tea is sold in the Salzburg region, you initiate an interactive query process, after which an ad hoc analysis is performed; this analysis forms the basis for deciding, for instance, whether to rent a warehouse near the border due to location considerations (see also Key Performance Indicators – KPI)
OLAP and Its Disadvantages in Logistics
Querying via OLAP from different analytical perspectives is certainly helpful, because, for example, a product manager is interested in different metrics than a division head. However, when examining the available metrics within a warehouse management system, OLAP generally accesses data records that have already been combined or at least consolidated from multiple sources. In addition, a data warehouse provides information in various formats. Consequently, if the analysis itself is complex and large amounts of data are required, databases can be tied up for extended periods (see “Data Mart” above). This is one reason why such analyses are usually run locally.
The OLAP Cube
The structure that illustrates OLAP is typically visualized using the OLAP cube. In this multidimensional cube, data is organized and stored as elements, and its dimensions (properties) represent the respective characteristics of a query; these, in turn, can include axes such as sales volume, net revenue, price, product, or region. Within this cube, the required key metrics are referred to as facts, which condense and reflect quantitatively measurable facts in a concentrated form. They are defined within specific values at so-called nodes (see figure), thereby enabling various basic operations through which data can be analyzed from different perspectives. In other words, by selecting a specific perspective or level of detail, reports and analyses with varying levels of insight can be generated. Criteria can also be combined with one another in almost any way*.
Summary
OLAP systems support decision-making within a company by analyzing and evaluating existing data from various perspectives. The questions posed to the system are derived from a hypothesis that is either confirmed or refuted. In this context, the data processing workflow often forms the technological foundation for business intelligence applications.
* Big Data Insider – What Is an OLAP Cube
Teaser image: Nick Youngson / CC BY-SA 3.0
IT Knowledge: Basics OLAP Cube
