The work of OLAP systems is based on a multidimensional data model; that is, such systems allow you to analyze many different parameters from different angles. They process multidimensional data arrays, that is, those in which each element of the array is associated with other elements.
Therefore, OLAP allows you to build hypotheses, identify cause-and-effect relationships between different parameters, and model the system’s behavior during changes.
At the same time, the data is organized in the form of multidimensional cubes – the monitored parameters will be the axes, and the data is located at their intersection. Users can select the desired parameters and get information on different measurements.
For example, for sales, the axes of the cube could be products, customer type, region, purchase frequency, and so on. The user can get data about what products are bought more often in what areas or what kinds of buyers shop more often, or how many products are sold in each region in a month.
The OLAP system collects information from databases, ERP, CRM, and other sources and forms a multidimensional data array. In general, the OLAP structure looks like this:
- Data sources – relational or multidimensional databases, data warehouse.
- OLAP server that manages multidimensional data arrays
- Applications that generate reports, graphs, and charts for users
How Can OLAP Be Implemented In Practice: Types Of Such Systems
The most straightforward and most obvious approach is to create a system that does not directly store anything but can quickly take out different records from different places and show the data in the correct form to managers. Such systems work well when the data is decomposed into the same DBMS type. For example, all divisions sit on a PostgreSQL relational DBMS.
OLAP with this architecture will be called Relational OLAP (ROLAP) – OLAP, built on the relationship of tables and databases to each other. Such a system does not require preliminary preparation of records in tables for analysis – you can take all the necessary values \u200b\u200b directly and online.
If the data is not only in the same type of corporate database, it is necessary to collect information from different sources and bring it all together. There is a stage of preliminary data preparation on a separate server. And such a system is already Multidimensional OLAP (MOLAP), or multidimensional OLAP. It’s more challenging to build such a thing, but sometimes you can’t do without it – the more significant your company, the more heterogeneous data storage systems will be involved. This is the most efficient type for analytical processing, as it allows you to structure data for different user requests.
And the third type is a hybrid of the first two types of systems. In very, very large companies, some of the data is easier to get through database queries, and some must be pre-prepared using multidimensional OLAP that works with various sources.
