A data mart is a small data warehouse used to support a specific business department function. It is focused on a specific subject or subject of a data warehouse. A data warehouse contains information about multiple subjects and takes much longer to build than a data mart. Because they use fewer sources and are single subject, they rarely exceed 100 GB.
Cloud Data Marts
Cloud-based data marts are becoming the norm due to the ability of a line of business to create them as they are needed without the IT administration and procurement overheads that traditional on-premise require. A line of business can deploy and populate a cloud version in days to model and explore new business opportunities as needed and only pay subscription charges for the short time they need to run ad hoc analysis projects. Cloud-based ones are also economical for long-term projects due to low ongoing administration costs.
Getting data into a data mart can be challenging. Businesses often struggle with maintaining many point-to-point integrations. Enterprise-level data integration tools can be expensive and too complex for departmental use. The Actian Data Platforms has built-in data integration to simplify populating your data warehouse. If the data mart uses data that is rarely accessed and is already stored in a structured format, such as a folder of .CSV files, it is possible with The Actian Data Platform to simply access it in place without loading it.
The following is a list of the most common reasons businesses use them:
- Easier to setup than a data warehouse because it involves smaller data sets.
- Faster deployment compared to an IT-led data warehouse project.
- Department-level autonomy allows a line of business to control data sources and use.
- Subject-specific data can deliver better performance than accessing a more extensive data warehouse.
- Lower cost to create as it avoids a capital procurement cycle when cloud-based.
Actian Data Management
A single Actian Data Platform subscription can support multiple data mart projects. Built-in integrations make populating the data marts easy. Learn more about The Actian Data Platform.
Data Mart Types
A data mart can be a subject-specific subset of a data warehouse, which is considered dependent on its parent data warehouse. It can be independent of a data warehouse when it contains data sourced from department-specific data sources to be used autonomously. A hybrid data mart contains a mix of data from a central data warehouse along with independent data.
Unlike transaction-oriented databases, a data mart is optimized for query processing. The database design often uses a star-schema organization consisting of a central fact-table and several dimension tables that enable queries optimized for commonly asked questions. Snowflake schemas don’t use well-defined fact tables but lower storage costs. This benefit is often offset by higher maintenance costs due to the added complexity of the structure when compared to a star schema. The low cloud storage cost means the simpler star-schema is by far the most popular data mart structure.
It is not uncommon for them to use in-memory databases or pre-aggregated multi-dimensional cubes to speed access to data that is changed infrequently, nightly, for example. Modern columnar database storage used by the Actian Data Platform provides high performance by using in-chip parallelization without worrying about pre-aggregations and complex cubes and indexes.
Sales Region Data Mart Example
A large corporation may have a global data warehouse for the Sales function. For example, the business could be segmented into three sales regions for the Americas, Asia, and Europe. At the corporate level, executives need a global view with summaries that drill down to the country level. This data warehouse lacks the detail the regional operations need to be successful. A data mart focused on regional sales can contain sufficient sales detail to allow area sales managers to adapt to local customer buying patterns.
Marketing Data Mart
The marketing function in any business needs to understand how prospects move through the sales funnel in detail, so outreach informs and guides the buyer’s journey. A marketing data mart needs to track the sources of leads from the first Google search through conversion and retention of customers after the sale. It uses data from sources, including weblogs to see what pages prospects visit, marketing automation data to see which emails got opened and clicked, Salesforce data to coordinate actions with Sales teams, and intent-based marketing systems to inform the next steps.
Retail Data Mart
Logistics operations span the globe. Tracking the supply chain of goods from farms or factories requires large data warehouses. However, local analytics at a destination retailer can be managed by a local data mart. A store manager can use it to track inventory levels, replenish inventory at hand, and monitor buying trends. These functions, aided by a data mart, allow the store to satisfy customer demands and remain profitable.