Analysts are not employed by SDxCentral and the views, thoughts, and opinions expressed in their content belong solely to the author and do not reflect the views of SDxCentral. Note: AvidThink is a separate organization, created by Roy Chua, that is not affiliated with SDxCentral.
The last week has seen a large number of news articles and blog posts covering MWC 2019 from the halls of the Fira in Barcelona. As I eyeballed most of the major technology publications, it seems that 5G was featured in at least 80 percent of the coverage. After a week on the show floor being bombarded by inane marketing messages and demonstrations, many barely even related to 5G, that was the last thing I was looking for online. Having said that, I believe we have a new contender for the next overhyped term as we enter 2019 and head towards 2020 — “edge.”
Yes, I believe edge computing will be the new hottest term, beating out artificial intelligence (AI) (but only barely), IoT, SD-WAN, SDN, and NFV. From hardware manufacturers like Dell, Lenovo, Hewlett Packard Enterprise (HPE), Kontron, and QCT to software and platform providers like Ericsson, Nokia, Huawei, Intel, Packet, Vapor IO, Arm, and Marvell/Cavium, edge was the term that appeared across all their booths. Unlike AI, which was certainly a hot topic but was limited to just software or platform providers, edge proved to be universal.
And yet, most of the booths presented edge as if it embodied a single truth. As with SD-WAN, which has been diluted to the point where a common definition is hard to nail down, it seems the edge will suffer the same fate.
At MWC Barcelona, many booths had edge stories: the deep-edge, the far-edge, the near-edge. The reality being that there won’t be just one edge. Instead, we’ll see a continuum from the cloud data center to points-of-presence (POPs) and central offices, to the enterprise WAN edge in remote locations, to gateway or mini data centers deployed in various locations (stadiums, multi-tenant dwellings, etc.), even to base stations in the radio network.
Each one of these locations will have different needs and requirements, necessitating a new generation of hardware capabilities and form factors as myriad system manufacturers tout their varying offerings. Hardware vendors will proffer hardened, optimized, converged systems that are ideal for running edge services. But wait, what edge services? We haven’t quite figured out the software stack yet — despite what some vendors are claiming — and the application development platforms are just being formed.
Speaking to the early entrants in the edge market it’s clear the industry is at the beginning of its journey to the edge. How will we get our applications packaged to run at the edge? What application platform will we use at the edge? Is it really Kubernetes and containers? How will we ensure that the platforms are leaner than our previous dalliances with OpenStack in the telco core? Will something like Rancher’s k3s do the trick for small form factor?
And how will we orchestrate, manage, and monitor these multiple clusters, which will easily number in the hundreds or thousands, many in remote mini data centers?
At the very least we will need to come together as an industry to start defining the different edge locations and characterize what’s needed at each. Likewise, for many of our edge and 5G applications we probably should think long and hard about whether a function is appropriately run at an edge data center or on the user device. And yes, standards bodies and open source organizations like IEEE/IIC/OpenFog, ETSI MEC, and LF Edge are all hard at work, but there’s plenty of confusion all around.
Before we jump and make large investments, I’m hoping the business folks at telcos and edge-related companies carefully evaluate the rate of change of technology for the user, the cloud, and edge locations, that cost of replacement for devices at these locations, as well as understand the intricacies of IT-free edge deployments. There’s plenty of work for all of us to do before we have enough to adequately hype up the edge. I’m sure that we as an industry are up to the task, plus 5G as a term is near the peak of inflated expectations and will no doubt start declining soon — you know the end is near when you hear “5G E” — so we need a new cool moniker to hang our hat on for MWC 2020, and it’ll likely be the “edge”.
Image: Vapor IO’s edge micro-data center at Arm’s MWC Barcelona booth.