What is edge computing? The why (and where) of edge computing

The cost of moving data

In conventional data processing approaches like cloud computing, the data is transferred to the cloud via the internet, or in some cases via a wide area network. Generally speaking, this approach works well for many applications but has a cost in terms of bandwidth needs. In some cases, the cost comes with moving data out of the cloud (known as egress charges); in other cases, the cost is in terms of the time that it takes to move data into the cloud.  Processing data at the edge can offer an efficient system that results in lower latency while avoiding costs for bandwidth and data storage.

For many types of AI deployments, for example, businesses have realized the limitations of centralized cloud infrastructure, convincing them to shift to edge processing that lets them less worry about bandwidth and latency constraints. Networking technologies such as 5G in tandem with edge computing could enable many AI deployments in remote locations.

One of the key benefits of edge computing for AI is lower latency.  AI-based applications rely on high accuracy models, and a quicker data feedback loop can be used to improve the AI model accuracy. After using data, it can be discarded rather than stored, resulting in another cost-saving benefit.

This highlights another benefit of edge computing: data sovereignty. As data collected at the source is locally processed, this allows organizations to keep the data well protected within a defined geographic location as well as a specific, localized ecosystem.

Visualizing which edge to use

According to LF Edge, a group within the Linux Foundation working to define standards and technologies, edge computing is defined as “Distributed cloud computing, comprising multiple application components interconnected by a network.” The cloud edge and device edge allow the conceptualization of multiple steps.

[Image Credit: LF Edge]

The two main edge categories are the Service Provider Edge and the User Edge. Before moving towards the first main edge tier (the Service Provider Edge, or SP Edge), let us understand the centralized data center. These are facilities that house cloud-based computing services, offering economies of scale and flexibility that are not possible on devices.

The SP Edge takes the same idea and, at a smaller scale, provides services over the global fixed networking infrastructure, often consumed as a service. The SP Edge is more distributed than a centralized data center. Most of this is leveraged from fixed and private networks developed by the Communications Service Providers (CSPs), many of whom are deploying both edge services and 5G services.

The second top-level edge tier is the User Edge which is another step away from the centralized data center.  Compared to the SP Edge, the user edge represents a highly diverse combination of resources. “The closer that edge compute resources get to the physical world, the more constrained and specialized they become,” the LF Edge Foundation explains. While devices at this edge have less compute and storage capacity than the centralized cloud, there are more of them, and developers can take advantage of the security, privacy that local computing at the User Edge brings to the table.

As the data has grown with complex applications coming to the market and the need for automation in every space of day-to-day life, edge computing has several advantages over current cloud service architectures, and these advantages will enable rapid market growth. According to market research provided by IDC in 2022, “Worldwide spending on edge computing is expected to be $176 billion in 2022, an increase of 14.8% over 2021.”

The firm said that “Enterprise and service provider spending on hardware, software, and services for edge solutions is forecast to sustain this pace of growth through 2025 when spending will reach nearly $274 billion.”

Article Topics

5G  |  cloud computing  |  data center  |  edge computing  |  latency  |  LF Edge