Databases Reference
In-Depth Information
Enterprise Data Platform Ecosystem - BDW
and EDW
A true enterprise data platform should leverage the synergy of the two technology stacks
(i.e., BDW and EDW). Together they can provide capabilities to exploit petabyte-scale
preprocessing of structured data and unstructured data such as free-form text (e.g.,
customer comments, user feedback, or product complaints) and semi-structured data
such as blogs and click streams. Using map-reduce technologies, the unstructured data
and semi-structured data can be transformed to structured results, which can be further
fed into the EDW analysis components as new attributes of significance, for example,
by first searching for relevant words and concepts and then quantifying the results with
counts or other statistics that reveal patterns. These new results can then be combined
with other existing facts or dimensions in the EDW to enrich analytic capabilities.
As discussed earlier in this chapter, the goal of BDW is to provide a platform that
helps in generating insights by following a discovery type of approach. The EDW data
elements (the dimension entities, aggregated facts, enterprise relevant KPIs, metadata
information) can significantly enhance this discovery process and shorten the time-
to-insight cycle. The conformed dimension tables in EDW reflect a standardized view
of business critical entities within the enterprise; they serve as a single source of truth
by linking records across several data warehouse fact tables or data marts, they are
validated by master data management processes and hence are usually de-duplicated
and cleansed. In addition to the dimensional data, other important categories of data
within the EDW are hierarchies, metadata, taxonomies, and business rules. These
EDW data components may provide a useful business glossary and cross-reference
data dictionary during the discovery processes in the BDW. The EDW is also a valuable
source of standardized facts, dimensions, and KPIs; these enterprise data elements can
be effectively leveraged during the discovery process in the BDW. For example, during
the discovery process in BDW you may notice few anomalies in data; but when you
reference the metrics in EDW you will realize some of those anomalies are valid business
conditions.
Figure
5-4
illustrates the enterprise data platform consisting of hybrid architecture
where both the data platforms contribute toward developing an enriched enterprise
data store.
