Embedded Analytics: a system engineering approach

Embedded Analytics, is a buzz word which has been used extensively by everyone in the recent past. Lots of companies including SAP have written about it. Reading all these articles published by CTO’s of these organizations, I strongly feel that people have broadly limited the scope of Embedded Analytics in their definitions.

 

What is Embedded Analytics?

According to the community embedding analytics into a workflow which enables the business users to understand the trends with latest information on-demand is embedded analytics. To simplify, enterprise reporting, embedded into work flow with authentication is embedded analytics. Embedded analytics has been viewed as a real-time reporting tool. Organizations working on Embedded analytics have started building reporting IDES to embed into workflows. BI organizations also are taking the same approach to embed BI reporting into work flows.

 

Embedded Analytics re-defined

I would like to redefine the scope of embedded analytics before discussing the approach.Embedded analytics is a framework in which one can do real-time analytics combined with building automated decision support systems where ever possible, which would feed and change the business work-flows and decisions real-time and also do real-time reporting. This makes the embedded analytics framework as the heart of the business and all other subsystems to be built around this.

Requirement for a System Engineering Approach

Frameworks like these should be highly scalabe, availabe and robust. The Total Cost Of Ownership (TCO) of these frameworks must also be less. All these systems must be designed for web scale (hundreds of terabytes of data) with all the flexibility of doing multi-dimensional analysis. One of the easiest way to implement these would be to use a reliable RDBMS for persistency and multi-dimensional analysis. The TCO of these solutions are very high and also achieving web scale may not be possible.
Due to these requirements a framework for embedded analytics with requirements as defined above needs an engineering approach.

More about the approach in the next blog…..

Ramkishan B K, Co-Founder and VP - Analytix and Consulting, ISACGlobal