BDW
1. Define, and illustrate using a diagram, the following: Primary Data Warehouse and Data Mart. In this connection, explain the difference between ROLAP and MOLAP.
A Primary Data Warehouse is a central repository of a database of a complete organization. It holds multiple subject areas and very detailed information. A Data Mart is a subset or an aggregation of the data stored to a primary data warehouse. It often holds only one subject area – for example, a specific department, finance or sales. It may hold more summaried data, and is typically smaller than a warehouse because of its employment on a different grain. Figure 1.1 illustrates the difference between data mart and a primary data warehouse. Since the data mart typically holds one subject area, it is much smaller than a primary data warehouse. These data marts can be viewed as small, local data warehouses replicating the part of primary data warehouse which is required by a specific domain or department.
Figure 1.1 A data warehouse does not necessarily use a dimensional model, since it is partly normalized RDBMS, but data marts are multidimensional cubes. This connection gives arise to two concepts, ROLAP and MOLAP. ROLAP is an implementation based on a relational database, in our case which is a primary data warehouse, and MOLAP is an implementation based on a multidimensional database which are data marts. ROLAP tools use the relational database to access the data and generate SQL queries to calculate information, whereas MOLAP tools are designed to allow data analysis through the use of multidimensional data model, but it differs significantly from ROLAP because it requires storage of information in the cubes, which are known as data marts.
2. Compare one-, two-, and three-layer DW architectures. What benefit does the reconciled layer provide in the thee-layer architecture?
One Layer Architecture: The goal of a single layer architecture is to minimize the amount of data stored, which is
A Primary Data Warehouse is a central repository of a database of a complete organization. It holds multiple subject areas and very detailed information. A Data Mart is a subset or an aggregation of the data stored to a primary data warehouse. It often holds only one subject area – for example, a specific department, finance or sales. It may hold more summaried data, and is typically smaller than a warehouse because of its employment on a different grain. Figure 1.1 illustrates the difference between data mart and a primary data warehouse. Since the data mart typically holds one subject area, it is much smaller than a primary data warehouse. These data marts can be viewed as small, local data warehouses replicating the part of primary data warehouse which is required by a specific domain or department.
Figure 1.1 A data warehouse does not necessarily use a dimensional model, since it is partly normalized RDBMS, but data marts are multidimensional cubes. This connection gives arise to two concepts, ROLAP and MOLAP. ROLAP is an implementation based on a relational database, in our case which is a primary data warehouse, and MOLAP is an implementation based on a multidimensional database which are data marts. ROLAP tools use the relational database to access the data and generate SQL queries to calculate information, whereas MOLAP tools are designed to allow data analysis through the use of multidimensional data model, but it differs significantly from ROLAP because it requires storage of information in the cubes, which are known as data marts.
2. Compare one-, two-, and three-layer DW architectures. What benefit does the reconciled layer provide in the thee-layer architecture?
One Layer Architecture: The goal of a single layer architecture is to minimize the amount of data stored, which is
References: Textbook ‘Data Warehouse Design’ By Matteo Golfarelli and Stefano Rizzi