<< back Page 2 of 3 next >>

Building a Competitive Data Architecture, One Technology at a Time

The challenge in making this a reality is the level of data literacy among end users, stated Jordan Morrow, senior director of data and design management skills with Pluralsight. “The key with any data and analytics technology is not the technology itself but understanding whether people can adopt it correctly. The technology can be sound and amazing, but if the workforce is not confident in its data literacy skills, the adoption of those products can be quite low. In order for these technologies to succeed and work the way they should, organizations need to combine them with data literacy learning and strategies.”


Importantly, today’s—and tomorrow’s—data architectures need to support the increasing use of real-time capabilities. Enterprises must be responsive to customers as they specify and consume digital products or services and also be able to move data rapidly across internal processes. “Long gone are the days of receiving weekly, monthly, or quarterly reports; scrambling to react to them; then making decisions that may be outdated by the time they’re put into place,” said Eric Raab, CTO of KX. “Now, organizations can bring together real-time and historical data from across the business for analysis in the moment to drive faster, smarter decision making.”

When real-time data is coupled with historical data, “organizations can gain greater context and also build predictive models to be able to act in advance,” Raab continued. “Streaming analytics underpins this shift to a real-time decision-making process, and more businesses will employ this approach to turn their data into actionable insights.” In addition, streaming analytics will move more to the edge, as IoT devices proliferate and explode data management needs. “Edge devices will gain greater processing power and capacity, and software will take increasing advantage of that. Edge-captured data will be aggregated in the cloud, and decisions made by machine learning and AI will be available in real time to mobile devices,” Raab said.


The convergence of edge computing and AI means greater responsiveness close to end users and the data they are generating or consuming. Edge, boosted by machine learning, “enables enterprises to process massive amounts of data locally, thereby reducing reliance on cloud networks,” said Jeffrey Ricker, CEO of Hivecell. “This makes data processing significantly more efficient, especially for businesses with hundreds or thousands of locations. Machine learning at the edge also addresses security concerns by processing sensitive data locally, rather than in the cloud.”

While AI-driven edge computing will reduce traffic from the cloud, it also will help manage the proliferation of 5G networks, Ricker pointed out. 5G—with greater capacity and intelligence built in—“could easily overwhelm the fiber networks, the data centers, and the cloud,” he explained. “The answer is to move compute power from the data center to the base of the tower to handle caching, preprocessing, and local processing. We are just scratching the surface when it comes to edge computing.”


The rise of pervasive and abundant data means a new generation of applications and use cases that more aggressively employ AI and machine learning. While AI and machine learning have been around for decades, “it is the availability of large datasets that have been created in recent years that makes the practical use and application of AI and machine learning productive today,” said Marshall Choy, vice president of product at SambaNova Systems. “As organizations achieve greater degrees of digital maturity and AI fluency, we are seeing an increase in project scopes and a shift in focus from cost reduction to gaining efficiency to drive profits.”

Looking ahead, Choy sees AI and machine learning “being pervasive across the enterprise as the backbone for all applications that are centered around use cases such as natural language processing, computer vision, and recommendation. AI and ML will transform business and technology in a more dramatic way than the internet did decades ago—refactoring the Fortune 500 and enabling capabilities that were unthinkable only a short few years ago.”

Big data is making AI possible, and AI is helping to manage big data. This “applied AI” is providing enterprises with  “the ability to harness structured and unstructured data at scale from both within and outside and derive descriptive, predictive, and cognitive insights,” according to Sreedhar Bhagavatheeswaran, senior vice president and global head of digital business for Mindtree. Applied AI is on its way to “becoming mainstream amongst most enterprises over the next 5 years,” he added. Use cases include “effective use of natural language processing techniques to enhance text-to-speech and speech-to text-technologies.” Bhagavatheeswaran also sees computer vision on the rise, in addition to classical AI/ML use cases powered by advances in deep neural networks, which will become an essential tool for making better business decisions.

<< back Page 2 of 3 next >>


Subscribe to Big Data Quarterly E-Edition