Page 1 of 3 next >>

Data Moves Closer to the Edge, And it's a Good Thing

The technology pendulum has long been swinging between centralization versus decentralization. Is it better to have a mainframe or distributed systems and data across Windows, Linux, and Unix platforms? Is it better to have these server farms or consolidate things with a cloud provider? Now, the action is moving away from the cloud, toward edge and Internet of Things (IoT) configurations. The enterprise market for edge computing will grow by 22% over the coming year, mainly for hardware, according to estimates from Deloitte Global. By contrast, enterprise networking equipment expenditures will grow by 4%, while overall enterprise IT spending is projected to rise by 6%.

There’s a good reason for the surge of interest—and corporate budget spending—to get to the edge. “We are seeing a trend in more computing shifting to the edge, especially as edge nodes become more powerful,” said Grace Lewis, principal researcher at Carnegie Mellon University’s Software Engineering Institute. “Processing data close to where it is generated, as opposed to having to send all the data to the cloud for processing and wait for a response, is a huge driver, especially in situations where support for rapid decision making is a requirement.” Edge computing “meets demands of critical business processes in a rapidly changing climate,” said Steve Currie, VP and distinguished engineer at Kyndryl. “It’s closer to the point of data generation, increasing the agility and responsiveness. Edge computing has been used in improving manufacturing quality control and maintenance, worker safety, retail loss prevention, customer buying pattern recognition, and warehouse logistics. Edge computing applies an optimized set of resources—compute, storage, network, and location—to the task at hand.”

Aron Brand, CTO of CTERA, offered up an apt analogy: “The shift of computing toward the edge can be likened to the decentralization of a city’s resources. Instead of having all essential services located in the city center, causing congestion and inefficiencies, they are distributed to various neighborhoods to allow for faster access and improved services. In the same vein, edge computing brings data processing closer to its source, leading to reduced latency, better performance, and an overall improvement in user experience.”

Edge computing inherently offers greater “security, privacy, cost reasons, as well as the need for low latency results,” said Vid Jain, CEO of Wallaroo. “In some cases, it makes much more sense to move the compute near the data, rather than the data to a central compute location.”

In its wake, edge computing is spurring many new types of applications. “We are seeing customers using compute resources like traditional servers processing images through small-form compute devices within vehicles,” said Currie. “Computing at the edge is occurring in manufacturing plants, retailers, mines, agriculture, ships, trucking, and more.”

A compelling example of data and computing at the edge is video surveillance, Fredrik Nilsson, VP of the Americas at Axis Communications, illustrated. “Cameras have been steadily improving in both image quality and processing speed, making it increasingly possible to process and analyze video at the network edge. The emergence of deep learning capabilities has enabled advanced, new analytics capabilities to be run on the edge devices themselves, allowing users to save on both bandwidth needs and video storage costs.”

There are numerous advantages to edge computing, starting with “reduced latency, because data processing happens close to where it originates or is needed and avoids round-trip time to the cloud,” said Lewis. In addition, edge and IoT networks help “reduce bandwidth use due to optimization of communication to and from the cloud.” Furthermore, edge offers “improved privacy and security because sensitive data stays local and enables the continuity of operations (in some cases) despite lack of connectivity to the cloud.”

At the same time, she noted that “security and privacy also become more challenging because of the need to protect off-premise edge nodes and IoT devices.” This, she said, makes security a mixed bag.

Edge computing continues to evolve, driven by advances in computing and sensors. The compute hardware side is seeing “an ever-shrinking size and power requirements with concurrent increases in compute performance,” said Mark Wright, director of product marketing, associative processors, at GSI Technology. “Edge itself has divided into multiple categories. While once it was customer premise equipment (CPE) and an interim stage of fog computing before enterprise processing, now we have extreme edge.” Examples of extreme edge include phones, cameras, doorbells, aggregators, set-top boxes, CPE hubs, edge servers, and content delivery networks, Wright added.


Demand for real-time capabilities is also accelerating the movement to edge and IoT computing. “As enterprises continue to add new applications and collect even more data, many are adopting edge computing strategies, allowing them to put real-time information in distant locations without compromising speed and accuracy,” said Anshuman Kanwar, SVP of technology at Reltio. “Edge offers real-time processing for applications that require it, such as autonomous vehicles.”

Edge computing “reduces latency for applications and users,” said Wright. “Some data only has a very short useful lifetime. Sending this data to a cloud and waiting for a response to return would be a nonstarter for certain applications. A simple hypothetical example is measuring traffic density from different directions to automate traffic lights. Sending this data to a cloud and getting a response would make the information stale and not of use for controlling traffic lights. It would also congest local internet lines. …”

A key benefit of edge and IoT “is the improved response time, as data processing is done closer to its source, resulting in faster responses and lower latency,” said Brand. “Another advantage is offline availability ability; edge computing allows devices to operate autonomously during connectivity or cloud service outages, ensuring that services remain uninterrupted. Lastly, edge computing provides improved security and privacy by minimizing the need to send sensitive information to the cloud, reducing security risks and enhancing privacy.”

Page 1 of 3 next >>


Subscribe to Big Data Quarterly E-Edition