Sign up for our newsletter and get the latest big data news and analysis.

Creating Effective Enterprise Data Architectures

Medidata_headshotIn this special guest feature, Moises Do Nascimento of Medidata discusses his 3-pronged approach to creating effective data architectures that have served him well, especially now that he deals with complicated, unstructured healthcare data. Moises Do Nascimento is Chief Data Architect at Medidata, a company that offers cloud software to help make medical clinical trials faster, cheaper and safer. Before arriving at Medidata, Moises spent over 5 years at Pay-Pal, learning to make large data sets useful in ways that drove real business results. 

Information can never exist without data. Having data does not mean you have information, or better said, having data does not mean you have accurate and timely information. Therefore, just as a business needs to be well-designed to function, so does data.

Since the introduction of Google’s BigTable, data technology is being redefined. Data is flowing into companies in a velocity never seen before. For example, as the chief architect at Medidata Solutions, I must think about the volumes and complexity of data created in clinical trials and how we can help our customers extract real insight from Big Data. With so many moving pieces, how can companies create effective enterprise data architecture while moving business at full speed?

Before we answer, let’s define effective. The objective of a good data centric design, beyond transforming data into information, is to extract value out of the data as it flows through the systems. During this era of real-time analytics, to be effective requires accuracy and timeliness of the information generated. The only way to achieve this is by thinking how the data needs to be stored, processed and analyzed before we design systems.

Over the past couple of decades, I had the opportunity to work on enterprise data initiatives in several industries across several countries. I have observed some key commonalities among companies that were able to master data and increase their competitive advantage:

  1. Become a Data Centric Company

Data is not only a technical challenge; it is also an enterprise opportunity. Enterprise data architecture can become a great medium to a company’s excellence by connecting the business metrics needs to the technology design. In order to become a data-centric company, a combination of two approaches to data architecture is needed.

The first one is bottom-up, where we create a data-centric software development lifecycle (SDLC) by adding an enterprise-level conceptual data design before road map planning. The logical designs can then be federated into the product team to make sure an enterprise standard for data and its metadata is followed without losing the velocity of the engineering teams, especially if the company is running agile methodology.

The second approach is top down, with an enterprise-wide alignment around a data governance process with a systematic way to track data definitions into one single dictionary.

  1. Master and Integrate Your Small and Medium Data

Big Data technology offers a great solution to managing unstructured data and processing a massive amount of it in near real time. However, Big Data does not offer a better solution for structured data management, which is much more constrained by hierarchies, referential integrity and consistency. That subset of data, though small in size, is key to take full advantage of even bigger unstructured data sets.

Big Data platforms such as Hadoop will eventually mature to address some of these patterns. However, at this point in time, it is important to rationalize the data patterns to leverage the strength of each available technology. At Medidata we bring together information from disparate sources – clinical reports entered by investigators as well as Big Data sets from wearable devices. Regardless of their source and size, these data sets need to be able to reside on a single integrated platform.

Creating a master data management strategy along with enterprise data architecture programs is key for the company’s ability to extract information from structured and unstructured data sets in near real time.

  1. Build an Integrated Big Data Platform

With a process in place to manage and master data in the organization, creating a robust, secure and scalable big data platform becomes a natural evolution of the architecture. The clear understanding of how the data flows across systems and how the final data sets are stored works as a design input to serve data to applications, APIs, interactive analysis, reporting and data sciences.

In conclusion, it is important to understand that these three steps can be done in parallel while a company runs its business. Data is a living organism. As it flows through the company, it will evolve and change. The key is to have a good understanding of data management gaps and put an architecture roadmap in place. Each one of these components will connect with others to create a system where data not only flows and becomes information, but also builds a wiser organization. After all, our ultimate goal is not information, but wisdom.

 

Sign up for the free insideBIGDATA newsletter.

Leave a Comment

*

Resource Links: