This chapter is from the book
This chapter is from the book
This chapter is from the book
Establish Modern Analytics Architecture
As analytics has matured over the past 20 years, analytic architectures have gone through a substantial transition from standalone desktops to enterprise data warehouses to Big Data platforms such as Hadoop. High-performance computing environments, such as clusters and grids, which were once specialty environments, are becoming mainstream environments for analytics. This has created a hodgepodge hardware and software legacy in data centers across the globe. All of this has occurred while the cost of computing power has dramatically decreased and open source software has gone mainstream.
The paradigm shift underway is a movement toward building lean analytic architectures, as illustrated in Exhibit 1.1, based on simplicity and open standards that leverage commodity hardware and open source software to drive the costs out of the architecture, provide platform scalability, and leverage the latest innovation. This innovation supports the execution of thousands of computationally and data-intensive predictive models in production deployments by large user bases with differing analytic and service-level requirements. Building, managing, and supporting the ecosystems required to deliver on these requirements means incorporating many different hardware and software products—both open source and proprietary. Even within a single vendor, products oftentimes don’t integrate seamlessly due to generational changes in software and acquisitions. Lean analytic architectures use a proprietary hardware and software solution when the solution provides a unique value but insist that the solution has open interfaces that make it readily integrated to other solutions.
)
Exhibit 1.1 Modern Analytics Framework
The streamlining reduces the complex administration and maintenance costs while creating efficiencies for analyzing and deriving insights from the data.