Edge Learning: How to Get an IIOT Edge in 2018

by Brad Johnson, on Dec 14, 2017 12:56:28 PM

As 2017 draws to a close, the Swim team wanted to look forward to 2018 and the changes we’ll see in the IIOT technology landscape in the New Year. This year, major industry analysts like Gartner, Forrester, Frost & Sullivan, and many others have highlighted Edge Computing as an increasingly important piece the IIOT puzzle. Edge deployments benefit from significantly decreased latency of data delivery, while providing a distributed computing network which ensures more consistent uptime. Pairing Edge Computing technologies with other recent advancements in Machine Learning and streaming analytics can enable industrial enterprises to transform raw sensor data at the edge into real-time insights structured for use in IT and OT applications. In this post we’ll make some predictions about both the industrial IT and OT landscape in 2018, and explain why Edge Learning will see wide adoption in the New Year.


1. Edge Computing Will Dominate 2018

Early adopters have already realized the benefits of employing an edge strategy in their IIOT environments. But according to Bsquare's 2017 Annual IIoT Maturity Study, only 7% of enterprises have advanced to the edge analytics stage in their IIOT efforts. 2018 will see significant adoption of edge analytics technologies, as businesses look to further capitalize on data generated by their own operations. Edge analytics enables businesses to learn from their operations data faster, closing the feedback loop both in terms of latency and location.

2. Machine Learning Becomes Part of the OT Toolbox
The challenge with IIOT data is not in collecting it, but understanding it. As mentioned in an earlier blog post, Deloitte’s Smart Buildings: How IoT Technology Aims to Add Value for Real Estate Companies report states that 90 percent of all enterprise data is unstructured. Machine Learning is the answer for finding patterns and relevant insights hidden within “noisy” streaming industrial sensor data. The self-training nature of Machine Learning makes it ideal for turning raw, unstructured data into structured data that can be used by external applications.

3. Edge Computing Enables Real-time Machine Learning Apps

According to a Lightbend developer survey, 90% of their data processing workloads include a real-time component. However, these real-time components are often bottlenecked by other components of an application. Illustrating this point, an IDC Research survey found that 75% of respondents believe slow data has inhibited business opportunities, while 27% indicated untimely data has negatively affected their productivity/agility. Furthermore, 54% of respondents claimed that untimely data limited their overall operational efficiency. Survey respondents revealed that nearly two-thirds of data moved via ETL was at least five days old by the time it reached an analytics database, clearly useless for any real-time analytics purposes.
The distributed nature of Edge Computing enables data processing workloads to be processed in parallel, significantly improving latency for real-time applications. As more enterprises focus on incorporating real-time insights into their business applications, Edge Computing will play an increasingly critical role in application architectures.

Learn More
Learn how SWIM uses Edge Computing to deliver real-time edge data insights with millisecond latency for industrial and other real-time applications.

Topics:Machine LearningSWIM SoftwareEdge AnalyticsEdge Computing