Data warehouses are used to collect and qualify data from operational systems so that historical performance can be analyzed.
What is it?
A data warehouse is a repository for making operational data available for analysis. Typically, that data will be taken from a variety of business systems, cleaned and loaded into the warehouse, where it can be explored and analyzed.
What’s in for you?
Data warehouses are an important part of many business intelligence initiatives and are used to power executive dashboards, which provide regular updates on key business performance metrics. The data warehouse has frequently been the place to find operational performance data.
While many organizations are exploring options such as data lakes or data meshes, the data warehouse remains a useful system for monitoring historical performance data.
What are the trade offs?
Data warehouses pull data from operational systems, clean, and repackage that data for analysis. This is typically done as a nightly or weekly batch process and is ill-suited to real-time analysis — which is increasingly significant in the digital economy.
Many organizations had less than stellar experiences with data warehouse implementations. Traditional data warehouses often came with significant up-front data modeling and data integration code (ETL), which resulted in overbuilt warehouses that didn’t provide a return for years and were costly to maintain. They worried too much about data gathering and not enough about data use, resulting in bloated warehouses with unused but expensive data.
How is it being used?
Data warehouses are often used to populate business dashboards that give executives insights into company performance. They are well-suited to this type of historical analysis.
While the traditional big-bang approach to building data warehouses has fallen out of favor, organizations can still benefit from this technology by using a test-and-learn approach, where the data warehouse starts small and built incrementally.