Data Warehouse: The Building Blocks

A set of vocabulary flashcards covering the definitions, features, approaches, architectural types, and components of a Data Warehouse as described in the Chapter 1 lecture notes.

Data Warehouse

A subject oriented, integrated, nonvolatile, and time variant collection of data in support of management’s decisions.

Subject-Oriented Data

Data stored by real-world business subjects or events, not by applications, where all data sets relating to the same subject are tied together.

Integrated Data

Data from several operational systems and outside sources that has been standardized and cleaned of inconsistencies before storage.

Time-Variant Data

Historical snapshots of operational data over past and current periods that allow for analysis of the past, relation to the present, and future forecasts.

Nonvolatile Data

Data extracted from operational systems and outside sources that is transformed, integrated, and stored at specific intervals, remaining stable once inside the warehouse.

Data Granularity

The level of detail kept within a data warehouse; keeping data summarized at different levels for efficient query satisfaction.

Top-Down Approach

A method where data is stored at the lowest level of granularity based on a normalized data model in a centralized warehouse, which then feeds dependent data marts.

Bottom-Up Approach

A method where data marts are created first for specific business subjects based on a dimensional data model and then joined by conforming dimensions.

Supermarts

Carefully architected data marts implemented one at a time as part of a practical approach to building an overall corporate data warehouse.

Centralized Data Warehouse

An architectural type where atomic level normalized data is stored in the third normal form without the use of separate data marts.

Independent Data Marts

An architectural type where data marts are independent of one another, often leading to inconsistent data definitions and standards.

Federated

An architectural type without one overall data warehouse, where data is physically or logically integrated through shared key fields and global metadata.

Hub-and-Spoke

An architectural type where atomic data in third normal form is stored in a centralized warehouse, which then provides data to dependent data marts.

Data-Mart Bus

An architectural type where logically integrated supermarts are built by conforming dimensions to provide an enterprise view of data.

Production Data

Source data coming from the various operational systems of the enterprise, often characterized by disparity.

Internal Data

User-kept private spreadsheets, documents, customer profiles, and departmental databases that add complexity to integration.

Archived Data

Old data from legacy systems periodically stored in online databases, flat files on disk storage, or tape cartridges.

External Data

Information from sources outside the organization used by executives to spot industry trends and compare performance against other organizations.

Data Extraction

The function of retrieving data from numerous sources, often into a separate physical environment to facilitate moving it into the data warehouse.

Data Transformation

The process of standardizing data types, resolving synonyms and homonyms, purging non-useful data, and sorting/merging data from different sources.

Synonyms

The occurrence where two or more terms from different source systems mean the same thing.

Homonyms

The occurrence where a single term means many different things in different source systems.

Data Loading

The function consisting of initial loading when the warehouse goes live and ongoing incremental data revisions.

MDDBs

Multidimensional database management systems used to store aggregated and summary data extracted from the data warehouse.

Metadata

Data about the data in the data warehouse, acting as the directory of contents and the glue connecting all parts.

Operational Metadata

Metadata used to tie delivered information back to the original source data sets, records, and coding schemes.

Extraction and Transformation Metadata

Metadata containing data about extraction frequencies, methods, and business rules for data extraction.

End-User Metadata

The navigational map of the data warehouse that enables users to find information.

Management and Control Component

The component that coordinates services and activities within the data warehouse by interacting with the metadata component.