With the advancement of IoT, customers not only can track their shipments, they can request specific information about the conditions the shipment experiences, said Jennings. “For example, the Pfizer coronavirus vaccine needs to be stored at ultra-low temperatures. Real-time logistics technology allows the temperature of the shipment to be measured and tracked throughout its entire journey. The customer can monitor the progress to ensure quality control. Shipping devices can also track whether a container has been opened or fallen over during shipment. All that information, with the use of IoT, can now be tracked.”
Challenges Ahead
The availability of a wide swath of technology—even sophisticated cloud services—does not mean the world is ready to go to real time in a big way, however. “Today, deep learning deployments are very limited and are primarily optimized for the cloud,” said Eli David, co-founder and CTO of DeepCube. And, even in these cases, there are extensive processing costs, significant memory requirements, and high power costs due to intensive computing demands, David noted. “Additionally, ‘real time’ is limited due to the latency, bandwidth, and connectivity requirements associated with sending data to the cloud and back for processing,” he said. “Cloud connectivity does not enable this level of real-time decision making, but many of these challenges also plague deep learning deployments on edge devices, including drones, mobile devices, security cameras, agricultural robots, medical diagnostic tools, and more, where the current size and speed of deep neural networks has limited their potential. Progress is being made, but the industry needs a solution that addresses challenges around the size and speed of deep learning models without sacrificing accuracy for sensitive edge device applications.”
Ahmed agreed that the journey to becoming a real-time enterprise is a long-term effort. “Most sophisticated organizations are already capable of supporting real-time technology. However, middle-sized and smaller organizations—especially those with budgetary and technical investment deficiencies—have some catching up to do. These organizations may still be relying on end-of-the day processing, rather than real-time and semi-automated data transfer protocols. With the advent of cloud-based systems, this issue may be alleviated somewhat.”
Of course, data is at the heart of it all. In manufacturing, for example, “almost all traditional industrial applications demand real-time data,” said Keith Flynn, senior director of product management for AIoT Solutions at AspenTech. “While the process still needs to run with acceptable quality, safety and sustainability are more important than ever. There is a shift to have access to smarter software that can make decisions without human interaction using embedded AI. These AI-enabled applications make it increasingly important to connect more with real-time sensors. What we need is more real-time data [that is] faster and more secure since today’s modern software can exist outside of the plant.”
In addition, “subject matter experts diagnose situations and issues [that are] entirely different and have a different understanding of what’s happening,” Flynn continued. “With embedded AI in applications, today’s data needs to be more robust and validated. When we rely on modern software applications to provide instructions on how to run a plant, we need to ensure the data powering those decisions are reliable. Likewise, the more sensors feeding data as inputs, the more variables need to be considered when making decisions.”
Real-Time Steps
A holistic understanding of the overall enterprise data architecture is key to enabling a real-time enterprise. Industry experts make the following recommendations to moving forward with real-time capabilities:
Put tried-and-true best practices into action. “Similar to the discipline required to determine how fast your storage or backup solution could perform 20 years ago,” said Loeppke, noting, “You need to understand all the components required to accomplish the task and make sure you are matching those components to your technology capabilities. The old adage that ‘you are only as fast as your slowest component’ still applies.”