A Prudent Business Analytic platform

Analytics today are vital to triumph in any business functions and it’s no wonder acquiring or building Business Analytics platform seems to be today’s mantra for many Enterprises. 

However to build analytics platform one has to truly understand and answer the broader questions like  a) how do we accomplish the analysis, b) how decisions should be made and c) how technologies should be evaluated and implemented to avoid half baked solutions with its negative side effects.

These side effects can be minimized if enterprises were to initiate an assessment to understand their current platform maturity levels. One way is to view how much of their current operations is supported by different types of analytics namely Descriptive (What happened), Diagnostic(Why happened), Predictive (What will happen) and Prescriptive (How to prevent). These can be categorized further from value add perspective as providing Hindsight (Observation), Insight (Comprehension) and Foresight (Prudence) in the order of magnitude of complexities as well as steps towards higher maturity progression.

My aim in this blog is to emphasize that enterprises have to take a holistic approach in building or extending Analytics for their Data Warehouses or Native Repositories first and create Data Management Solutions for Analytics that utilizes both internal and external data.

There are many technologies both in hardware and software that implementer's can choose from with various degrees of mixes to provide an  enterprise level analytics platform. Some companies prefer to align with single vendor to obtain BOB (Best of Breed) products across their entire stack thus avoiding the fear of difficulties of having to bring and maintain different skills sets while others prefer to go with a different approach of getting the Best of BOB in the market from across vendors to integrate these products and technologies and build seamless platform.

Both these extreme positions or any varied combinations have to still overcome many integration issues with their enterprise data while providing plumbing activities (ETL/ELT) for data movements before it can be build as a cohesive and a high performance dominant platform. We also hear that many platforms sooner than later, plunge both on storage capacity as well as on performance with storage capacities overshooting their estimates and missing SLA's (Service Level Agreements), benchmarks on business users usage demand and requests.

On the horizon towards a Solution

Today we hear implementer's are vying with a concept what Gartner calls it as “Best Fit Engineering” (BFE). In this mix, the minimum required technology for each function is considered for an appropriate purpose and is therefore much more likely to exhibit a lower cost and still retain good performance.  Some of the areas that I believe are major areas where implementer's are considering are as follows:

Data Virtualization layer (DVL)

Implementing Data Virtualization layer over data stack creates a LDW (Logical Data Warehouse). This LDW would minimize the most expensive data movements (ETL/ELT) and expose the data requirements from different silos large and small by creating a repository of metadata layer to meet the data demands in format, scope across different channels.

Big Data Platform

Big Data platform is no more a buzz word or a hype nor an aspiration but a certainly a catalyst in providing many beneficial use cases for data analytics in performing exploration by utilizing it as ‘sandboxes’ for offloaded history from warehouses and external data.

Data Lakes

Data lakes can be deployed, managed and scaled for its computing needs and storage by building in public or private clouds for both external and internal data. Data lakes provides the much needed agility in terms of responsiveness and flexibility to deliver faster insights. The other big advantage in Data lakes is it does not restrict to pre-definition of schemas and can be propagated from different silos in their original formats. Organizations can utilize the BI Self-Serve tools much more effectively to prepare data for analysis in Data lakes while aligning to business needs and provide evidence based data discovery to support decision making.

Data Sciences

Build Data Sciences as a discipline within Organizations and nurture data scientists who can play pivotal role in sharing data science discoveries while processing operational applications data in the enterprise planning efforts and its executions. Nurturing and supporting this step could lead in creating building blocks for Graph databases that could also abridge gap for its usage between commonly used BI solutions by less skilled personnel.

Observation

There were many disruptive technologies we have seen in past and at each juncture it helped the IT folks to build faster, modular yet powerful and flexible platform and applications, however in the BI space the shift of balance of power is moving from IT to Business folks with ambitious Self Serve Service platform offerings.