Let’s say there is a supermarket and it is equipped with a few tens of cash register. All of them are connected to a database server and information about all the purchases is saved to this database in real-time.
It works just fine, but we need to analyze this data to make some reports or predictions. Such a database could take Gigabytes and it is impossible to run complicated queries over this data just because all the cash registers will freeze till the query is completed.
And you can guess that analysts also don’t want to work during nighttime when the shop is closed. They’d better come to office from 9AM till 6PM.
This is what Data Warehouse is for. It allows to make a snapshot from the database and build an OLAP cube. This cube could be used by analysts for complicated queries. It is located in a separate server. Even if query will take really long time, it will not impact on cash registers operations.
In fact for analysis it is not important to have the freshest data. It will be fine to update the snapshot (OLAP cube) on weekly or monthly basis.
Examples of Data Warehouse products:
- SAP BW
- Oracle Data Warehouse
- Amazon RedShift
- IBM has a set of industry-specific solutions