Page 88 - Building Big Data Applications

Chapter 3 Building big data applications 83

The new data layer needs to be harnessed with machine learning algorithms and an
artificial intelligence data munging and processing layer. This layer is where the intel-
lectual property and enterprise-specific analytic intelligence can be built and deployed.
This layer also will be crucial in securing the enterprise story of information manage-
ment as IT will don the role of the facilitator and not the owner of the processing layers
which will mean budgetary shifts and controls in the entire processing engine of the
enterprise. This is what I refer to as the impact of data science.
Data science is not new but rather has become a very important method to focus for
delivery of big data processing and management of all applications related to big data.
The concept further needs to be standardized as each enterprise is implementing its own
version of a data science team. The role of a data scientist is to become an explorer, a
researcher, an analytic nomad, an enthusiast of data discovery, and a deep sea diver.
Each of these behaviors have traits associated with them and these traits will be the key
to what the business wants to achieve in the world of information management and
applications. These newer roles will also define the governance model of execution as
business subject matter experts will not only define the rules but also enforce the rules
successfully. In this model of governance is where the risks of big data application are
managed effectively and succinctly.
The best-in-class companies that have experienced success have repeatedly said that
shifting the gears of application building and management from the hands of IT has
been critical to the overall success. While this concept is not new, the confusion of who
owns the data in the enterprise and the stewardship of the data within the enterprise has
emerged to be an intense battle of wits and courage. The biggest risk is not identifying
the owners of the layers of information, and not defining the models of governance
which will lead you to failures and disappointment with this new world of data and
applications.
Further additions to data science have emerged in the world of information man-
agement with robotic process automation and low code models of user interface
development. These new additions provide devOps teams benefits including ability to
develop in Python for backend processing, the ability to add frameworks for simplifying
the data collection and tagging exercises, the ability to create user interface layers in a
nonmodel-view-controller architecture, all of which provide applications layers
immense performance benefits.
No discussion is complete without touching security. In the new world of application
processing, security takes a whole new level of presence in the enterprise. The appli-
cations will be managed as services with isolation layers built to shield the actual layers
of data, which means even if there is a breach and compromises are made in the
application layer it will be detected, isolated, and managed without the data layer getting
affected and impacted. This is a discussion of how cyber security will work in the en-
terprise, and we will cover this in a chapter toward the end of this book.