Data Mart

Data Mart is "Subject or Application Oriented Business View of Warehouse". It is a subset of the information content of a data warehouse that is stored in its own database,summarized or in detail. It is a repository for data to be used as a source for business intelligence. It is not used as part of day-to-day operations. Instead, the data mart periodically receives data from the online transactional processing (OLTP) systems. The data in the data mart is then made available to Analysis Services for creating cubes with preprocessed aggregates. A data mart is made up of measures, dimensions organized in hierarchies, and attributes. Once data is organized, build the database structure for the data mart using either a star or snowflake schema.In simple terms, DataMart's are

• Quick Solution to a specific Business Problem
• can be used in Finance, Manufacturing, Sales etc.
• Smaller amount of data in datamarts can be used for Analytic Processing

Data mart has single or partial subject area such as customer.it has few source of data mostly restricted to one department.Implementation time for data mart is 4-12 months.it is restrictive, less extensible, has short life and project oriented. For e.g. the cost center hierarchy or account hierarchy details differ for different line of business (or department). Even the customer data may have some extra or different attributes for different departments. Data marts in different department will differ for customer data.
UDM (Unified Data Modelling) makes it possible for business intelligence to be extracted right from the OLTP systems in a manner that does not put undue stress on these systems, thus eliminating the need for data marts. But there are situations still exist where a data mart may be the best choice as a source for business intelligence data.

It is very difficult in practice to operationally support more than a handful of datamarts.so data warehousing comes into picture.Data warehouse is enterprise wide.It captures detailed data for the enterprise and is flexible used for ad-hoc query / analysis requirements. It has very few summary data. The data is kept in most granular form so it can be summarized later as per need.When using data marts,always design the data warehouse for full data retention,and consider storing data in archive until a requirement to analyze all the detailed data exists.
Source: Delivering Business Intelligence with Microsoft SQL Server 2012 (3rd Edition) Author: Brian Larson
.