Worldwide spending on edge computing is expected to see double-digit growth this year, according to new figures from analyst IDC.
It also predicted investments in edge will reach $176bn in 2022, an increase of 14.8 per cent over last year.
“Edge computing continues to gain momentum as digital-first organisations seek to innovate outside of the data centre,” IDC research vice president Dave McCarthy said in a statement, adding that the diverse needs of edge deployments have created a market opportunity for technology suppliers, increasingly through partnerships and alliances.
A great deal has been written about edge computing over the past several years, and yet it remains an area of technology that can be hard to define, as The Register has previously pointed out.
This has not stopped industry from trying to market edge solutions, so perhaps it is instructive to see what the industry currently thinks edge computing is all about.
IDC claimed to have identified more than 150 use cases for edge computing across various industries and domains, but says that the two edge use cases that will see the largest investments in 2022 are content delivery networks and virtual network functions.
In the enterprise space, the use cases IDC sees getting the largest investments in 2022 include manufacturing operations, production asset management, smart grids, freight monitoring, and intelligent transportation systems.
Process of elimination
The concept of edge computing is deceptively simple: put the processing where the data is being generated, rather than streaming it all back to a remote data centre, with the latency (and network costs) that inevitably incurs.
Step beyond that, and edge computing has such a diverse array of potential use cases that they often have little else in common, and thus building up a standard model or platform for “edge computing” remains elusive.
Meanwhile, one of the growing trends is for edge computing to be almost synonymous with AI, often because an application may call for a rapid or real-time response. Naturally, this model of edge computing is championed by Nvidia, which manufactures the GPUs that have become the accelerator of choice for many AI applications.
The GPU-flinger’s view of edge computing is actually quite pragmatic: it is the practice of moving compute power physically closer to where data is generated. The firm contrasts it with cloud, stating that edge computing reduces the need for large amounts of data to travel among servers, the cloud and devices or edge locations to get processed, which is particularly important for modern applications such as data science and AI.
Nvidia has its EGX platform designed especially for edge applications where data science and machine learning play a big part. This basically sees its GPUs married with Nvidia-certified systems hardware, operating a stack based on its CUDA framework and Kubernetes. In other words, it’s up to someone else to put it all together for you, Nvidia just provides the bits and pieces.
One firm that is getting more directly involved is Lenovo, which launched a new edge system, the ThinkEdge SE450, in December 2021. This is a compact ruggedised 2U server based on an Intel Xeon Platinum processor. It is also intended for applications involving AI and analytics with the ability to fit up to 4 GPUs.
At the launch, Lenovo revealed it is working with Barcelona City Council and local universities on projects that use its technology in “Smart City” style applications, using hardware such as the SE450 from its ThinkEdge series deployed in on-street cabinets dotted around the city. One application involves traffic management, analysing video feeds from cameras to detect if an accident has occurred and alerting the authorities.
One notable aspect of the ThinkEdge SE450 is its XClarity Controller, an embedded management engine that allows the system to be reached via a wired or wireless network connection for admin purposes, rather than having an engineer visit the site to perform maintenance. Managing edge systems is considered one of the major challenges of edge deployments, with kit distributed in remote locations rather than clustered together in a data centre.
Dell has similar systems, and also similar management features with the iDRAC9 management controller inside its servers. The firm outlined its view of the edge last year, in a blog by Alison Biers, director for Global Edge Solutions Marketing.
Biers said that Dell defines the edge as “where data is acted on near the point of creation to create immediate, essential value”, but goes on to add that delivering value at the edge requires the ability to consolidate and simplify IT and OT (operational technology). OT refers to systems used to monitor and control physical infrastructure, such as in building automation applications.
As well as ruggedised laptops, PowerEdge servers and its specialised Edge Gateway nodes, Dell also has the VxRail satellite node, which it introduced last year. As The Register noted at the time, this is a single-node version of the company’s hyperconverged platform which seems to be designed to allow a VMware environment to be operated in an edge scenario, but without the redundancy usually offered by a HCI cluster.
Perhaps the main issue with edge computing is that it is really a bunch of diverse applications that have been lumped together into one category, simply because they operate outside the bounds of the traditional data centre.
However you choose to define it, it looks like it isn’t going away. In Red Hat’s 2022 Global Tech Outlook report, edge computing was listed among the emerging technology workloads that organisations are most likely to consider over the coming year.
In fact, if you consider edge and IoT to overlap somewhat, the two combined were the leading category, with 61 per cent of respondents saying they were considering one or both.
Red Hat itself defines edge computing as a distributed computing model in which data is captured, stored, processed and analysed at or near the physical location where it is created. The firm says that it views it as “an opportunity to extend the open hybrid cloud all the way to the data sources and end users.”
This view that the edge is largely an extension of the cloud or the corporate data centre is shared by many vendors and analysts, with another recent report from IDC emphasising the need to find the appropriate location for workloads. Because workloads can reside across a continuum of core, edge, and endpoint locations, edge computing requires a significant amount of coordination among technology and service providers, it says.
IDC believes that the most significant edge workload opportunity is streamlining business intelligence and analytics, but perhaps surprisingly, does not see business application workloads as critical to the development of any major enterprise edge use cases.
“As edge technology continues to expand in usage in a variety of workplace environments, we are seeing growing interest in expected concurrent workload growth in areas such as business intelligence and analytics, AI/ML-related workloads, and content workloads,” IDC senior research analyst Max Pepper said in a statement announcing the report.
However, he added that the rapid deployment of edge computing is significantly shaping workload evolution.
Maybe the real lesson of edge computing is that an edge deployment will be intended to deliver a specific solution, and that this may demand specific hardware, software, and connectivity to meet those requirements, rather than just an off-the-shelf product.
In this case, the real opportunities for edge computing could lie with the systems integrators, which have the relevant skills to pull together a solution from various component parts and provide services to support customers in operating it.
This has been noted by edge vendor Stratus Solutions, which stated in a recent blog that the skillset of systems integrator engineers has never been in greater demand, and that they are extremely well positioned to benefit from the era of change.
Perhaps for this reason, IDC predicts that spending on professional and provisioned services will grow at a CAGR of 19.6 percent over the next five years. By 2025, it believes services will account for nearly 50 percent of all edge spending, led by investments in connectivity and edge-related infrastructure, platform, and software-as-a-service spending. ®