MIDSTREAM DaTA WAREHOUSE
We build your data warehouse using our pre-built midstream BI Accelerator models, modeled specifically for analytics, so that every decision-maker in your organization can have access to the data they want, anywhere, anytime. We follow the industry-proven and leading methodology of dimensional modeling inventor Ralph Kimball (www.kimballgroup.com ). Some key benefits from this methodology are:
- The BUS Architecture – It is extremely challenging for a BI manager to understand the content and location of the organization’s source data, while also understanding what keeps management awake at night. The manager takes refuge in carving off a little piece of the required solution for a given department and bringing it to completion. This creates stovepipe solutions that are dead ends that continue to perpetuate incompatible views of the enterprise. The BUS Architecture establishes a framework that guides the overall design, but divides the problem into bite-sized implementation chunks. So we can build out our integrated EDW iteratively, business process by business process, using a family of shared conformed dimensions to provide the required integration (between isolated pieces).
- Understandability is one of the primary reasons that dimensional modeling is the widely accepted best practice for structuring the data presented to the BI layer. Information in a dimensional model is grouped into coherent business categories (called dimensions) that make sense to business people.
- Extensibility – Dimensional models are gracefully extensible to accommodate unexpected new data. Graceful extensibility means that no query or BI application needs to be reprogrammed to accommodate the change.
- Query Performance is the other dominant driver for dimensional modeling. Denormalized dimension hierarchies and decode lookups can have a significant impact on query performance. The database engine leverages the star join by first constraining the dimension tables and then querying the fact tables with the Cartesian product of relevant dimension keys.