Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Supercharge IoT analytics with GPUs

Try going a whole day without the internet of things. In 2018, that’s almost impossible. Many of us wake up to IoT when our sleep trackers gently nudge us awake and give us an instant report on the quality of the last six to eight hours (on a good night!).

We get ready for work with an IoT toothbrush that assesses our brushing; we take the last iced coffee out of the fridge and it automatically reorders through Instacart; we leave the house and the smart lock kicks in when we exit our geofence; and the car automatically adjusts to our seat, temperature and musical preferences. All that and we haven’t even made it to work yet!

But the value of the IoT ecosystem is in the personalized digital relationships it can build, by correlating data across users, devices and connected things, and translating it into instant insight that makes our lives — or at least our mornings — better.

Businesses see the flipside. If they can analyze and act on IoT data when it’s most valuable, they can offer us products and services tailored to our wants and needs. To capitalize on this massive market opportunity, though, they can’t let their data sink to the bottom of the typical data lake. They need the right tools to analyze, understand and act on their data in real time.

Toyota, for example, recently went through a global reorganization to expand its work in data science technology. It launched Toyota Connected, focused on data-driven initiatives that include offerings like connected cars that share traffic details, telematics services that learn the customer’s preferences, and insurance models that price according to actual driving patterns. Now, Toyota will focus not just on producing great cars, but on the power of IoT data to reinvent the driving experience — making it more personal, safer and more appealing to customers throughout the life of the car.

IoT is likewise redefining industry, from manufacturing to utilities to logistics. For example, individual smart utility meters can deliver IoT data that helps the utility determine different rates for different seasons or times of day, and gives them the opportunity to offer consumers conservation incentives, as well. And with smart meters, utility companies can proactively change out overloaded and older equipment that shows evidence of potential failure or fire risk. It’s clear that the value of IoT is in building more intimate digital relationships by correlating data across users, devices and things, and translating it into instant insight.

Most businesses, however, are still using technologies that rely on the old serial computing paradigm, running on CPUs to store, manage and analyze IoT data. The problem is these technologies are just too slow to extract value from data in real time, to make operational decisions on the go, or to quickly and accurately assess risk.

CPU-powered databases take a long time and require users to decide what elements of the data they think will be important to analyze in advance, and they struggle to produce real-time visualizations. Traditional CPU-powered databases aren’t designed to handle the increasing complexity of both data sources and analysis; today’s data can be big or small, static or streaming, structured or unstructured, human or machine, long-lived or perishable. They’re holding the internet of things back.

IoT requires an insight engine powered by GPUs that analyze data simultaneously in real time. By using GPUs, a technology pioneered by Nvidia, companies can process data in parallel.

While a CPU is designed to process a trickling stream of data, a GPU is designed to process a rushing river of data — the difference between a hose and a waterfall. IoT is already generating 100x more data 100x faster than ever before, and requires a different foundation for success.

A GPU database can also take geospatial and streaming data and turn it into visualizations that reveal interesting patterns and hidden opportunities, whether that’s days, times and locations where traffic backups occur, or where it’s safe and profitable for an oil company to drill. It can also apply algorithms to augment human knowledge with artificial intelligence, quickly identifying complex patterns that are hard for humans to pick out, but which only a human can analyze in depth and in context.

At its core, with accelerated parallel computing, a GPU database makes it possible for businesses to process extreme volumes of IoT data — from the streaming to the historical — visualize it, analyze it and instantly feed insight back into the business for immediate action. Across industries, this will be the competitive edge in IoT.

All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.

CIO
Security
Networking
Data Center
Data Management
Close