Data Virtualization provide numerous opportunities for productivity gains and effort reduction especially in the case of using virtual data building blocks.
Creating and sharing virtual data building blocks is a terrific way to save your analysts’ time by reducing duplicative and redundant effort. These building blocks also serve to standardize how KPIs and metrics are calculated within your organization and provide the means for any analyst to use prepackaged datasets to streamline their efforts and ensure consistent data quality and accuracy.
The idea of the virtual data building blocks is to create different types of virtual views that can be stacked as needed to quickly serve new initiatives with minimal additional development time. There are three suggested types of virtual views in this approach: macro, filtering, and micro level virtual views.
At the macro level you would build broad re-usable views without filtering that can be used as a base layers for more specific needs. So, for example, for website traffic you might build a virtual view that contains all web traffic from all devices and for all purposes on all dates. It is unlikely you would ever need to query that view directly because of its broad scope but building it as a foundational layer provides a starting point for anyone wanting to analyze any aspect of traffic.
On top of the macro view layer you would then create filtering virtual views as needed. In our example of web traffic a filtering view might be all product browsing traffic. Another example might be purchase conversions, by landing page, by device. Filtering views can be easily re-used when similar specific data subsets are needed.
An initiative-specific micro view layer would sit on top of the filtering layer which sits on top of the macro layer. The micro layer joins across one or more filtering layers and can apply additional filters and aggregations so that only the desired data for a specific initiative is presented. This micro layer serves data visualization, data discovery, BI, and many other use cases precisely. These views can be shared and re-used but are less likely to have broad audiences like the Macro and Filtering layers will have.
Dirk Garner is a Principal Consultant at Garner Consulting providing data strategy consulting and full stack development. Dirk can be contacted via email: firstname.lastname@example.org or through LinkedIn: http://www.linkedin.com/in/dirkgarner