Home‎ > ‎Database‎ > ‎

Data mart

Data marts are stored data used for information in different departments of an organization.  Many companies start to build their data warehouses by building data marts first. Because they are in different departments they can be combined to create one large extensive data warehouses. Data marts are often subsets of data in a data warehouse. Some organizations have both data warehouses and data marts. That is when the data marts acts as subsets of the warehouses. They can enhance the functionality of larger data warehouses. Sometimes data marts are built because data warehouses are too large and difficult to create. Data marts are often separated based on the different departments within an organization. Each department usually owns their own software and hardware for data marts. Data marts can be built to support Point-of-Sale transactions. 


"...The data warehouse is nothing more than the union of all the data marts...," Ralph Kimball, December 29, 1997.

Star structure or star schema
The star design is formed when members of the department are required to follow certain guidelines set by the department. The Star is formed based on how an organization and its different departments are formed. So, if the organization is structured base on its functions than each function would have its own data mart, if the organization was structured on products then there would be a data mart for each product. The data mart only contains enough data in order to suit the needs of the department. This usually allows for flexibility, but does not contain large amounts of data. Star schemas are the simplest structure of data marts. 

Types of Data Marts

Dependent Data marts- source of data is from
data warehouses
Independent Data marts- source of data is from legacy applications environment. 

All data from dependent data marts are sourced from the data warehouses. The data from independent data marts uniquely comes from other areas, which is unlike dependent data marts. Dependent data marts are structured very well and independent data marts aren't structured as well. Independent data mart aren't built for long-term situations. Independent data marts need multiple data marts built in order for them to be effective. 

Data Marts Vs. Data Warehouses
Data Marts are easier to access than
Data Warehouses. Data Marts are cheaper to make than Data Warehouses. They are also easier to create than warehouses. Data marts focus on specific departments. Data Marts are generally cheaper and easier to make than warehouse, but sometimes they are complex and expensive to make. In a corporation, they may have one large Data warehouse and have a number of different Data marts for each department. So, if someone needed specific datafor a certain department they might go to a data mart and for data on the whole organization they might go to the data warehouse. Data marts are often used for more short-term success, whereas data warehouses may be used for more long-term success. Data marts are structurally different from data warehouses. Data marts usually contain summarized informations while data warehouses usually contain very detailed data. Data marts don't contain the same amount of history. Relationships and queries are different in data marts. Users in data mart are referred to as farmers.

Reasons to Build Data Marts
Sometimes data marts are built because of the technical limitations to build warehouses. Data marts could also be used to reduce the size of large data warehouses. If a business makes multiple products that are very different from each other then it may be necessary to build data marts. For security reasons a company might want to build a data mart. They would do this so that people in each department can access data thats only in their data mart rather than accessing data about the whole company. The fact that only the specific departments can see their own data can also be a problem too. In some cases other departments might need information from the data mart. If the business wanted a more extensive database then it probably would choose to build a warehouse instead.



Comments