Used to support forecasting and decision-making processes across the enterprise, a data warehouse acts as a centralized repository of an organization's data, ultimately providing a comprehensive and homogenized view of the organization.
Traditional database systems (e.g. SCT Banner, currently in use at Rensselaer) are designed to support typical day-to-day operations via individual user transactions (e.g. registering for a course, entering a financial transaction, etc.). Such systems are generally called operational or transactional systems.
A data warehouse complements an existing operational system and is therefore designed and subsequently used quite differently. While an operational system is transaction or process-oriented, a data warehouse is subject-oriented, geared toward flexible analytical processing of high volumes of business data.
what data exists within a data warehouse?
Large volumes of detailed data already exist in Banner and other transactional database systems within Rensselaer. A core subset of this data will be imported into the data warehouse, prioritized by subject area (i.e. by business area), including finance, research, contracts and grants, enrollment analysis, alumni, etc.
A fundamental axiom of the data warehouse is that the imported data is both read-only and non-volatile. As the amount of data within the data warehouse grows, the value of the data increases, allowing a user to perform longer-term analyses of the data.
Whereas the operational data is generally real-time or near real-time, data within the data warehouse is historical. The data import process described above will occur at specific intervals, likely once per day (during the middle of the night).
Such an import schedule should be sufficient since the data warehouse is used primarily for reporting and analyzing relatively large volumes of historical data in an effort to decide what to do in the future.