Implementation of Big Data Analytics - Big Data Analytics: Disruptive Technologies for Changing the Game - page 62

for analyzing the data and then feeding the results into a set of actions or reports.
Since all the data must be routed via a storage medium using a data warehouse,
the storage, organization, and retrieval of data creates a bottleneck. Typically,
the traditional approach requires a reorientation of the data from the data source
to a system of record and then into a set of models conducive to analytical
processing—which typically requires a number of data modelers, database
administrators, and ETL analysts to maintain the various data models and associ-
ated keys. Changes to the business environment require changes to models,
which cascade into changes across each component and require large mainten-
ance organizations.
Many components have already started to break off from this traditional
model. Netezza as the Data Analytics engine does not strictly follow this
paradigm, and it signiicantly reduces the model maintenance costs by reducing
the need for representation and key-driven performance tuning. Use of SPSS and
Cognos as user interfaces to drive modeling and reporting using Netezza's data
manipulation capabilities reduces the repopulation of data in analytics tools.
The revolutionary approach involves creating a brand-new Big Data
Analytics environment. We move all the data to the new environment, and all
reporting, modeling, and integration with business processes happens in the new
environment. This approach has been adopted by many greenield analytics-
driven organizations. They place their large storage in the Hadoop environment
and build an analytics engine on the top of that environment to perform orches-
tration. The conversation layer uses the orchestration layer and integrates the
results with customer-facing processes. The stored data can be analyzed using
Big Data tools. This approach has provided stunning performance but has
required high tooling costs and skills.
In a typical evolutionary approach, Big Data becomes an input to the current
BI platform. The data is accumulated and analyzed using structured and unstruc-
tured tools, and the results are sent to the data warehouse. Standard modeling
and reporting tools now have access to social media sentiments, usage records,
and other processed Big Data items. Typically, this approach requires sampling
and processing Big Data to shelve the warehouse from the massive volumes.
The evolutionary approach has been adopted by mature BI organizations. The
architecture has a low-cost entry threshold as well as minimal impact on the
BI organization, but it is not able to provide the signiicant enhancements seen
by the greenield operators. In most cases, the type of analysis and the overall
end-to-end velocity is limited by the BI environment.