Predicting the future is risky business. Predicting the future of network technologies is even more dangerous, mainly because our predictions are based on existing theories.
However, we could play devil’s advocate. For example, when someone says ChatGPT produces fantastic results, we could argue that ChatGPT repackages oodles of existing information — so all it does is massive plagiarism.
You see what I did there? Before you start typing up a lengthy rebuttal of my outlandish claim about ChatGPT, let’s play devil’s advocate with network-related technologies expected to thrive in 2023.
SEE: Don’t curb your enthusiasm: Trends and challenges in edge computing (TechRepublic)
- Why 2023 isn’t the year for multi-edge computing
- Why 2023 isn’t the year for private 5G networks
- Playing devil’s advocate on the edge
Why 2023 isn’t the year for multi-edge computing
Multi-access edge computing and 5G can work together to deliver new services. There is no question that MEC and 5G will reduce latencies and enable fast responses.
Most predictions claim edge computing will boom in 2023. That is up for debate. Edge computing is one of those technologies that is a hammer looking for a nail. Most MEC use cases like AR/VR, autonomous cars, healthcare and others are in their infancy. Some applications that require fast responses may move to the premises rather than the edge to avoid any dependency on the network.
For example, autonomous cars make immediate driving decisions based on what is happening in the car’s vicinity. They do not ask an edge node in the network to tell them to brake and not hit the vehicle slowing down in front of it. Amazon is moving towards speech processing in Alexa devices rather than depending on the network edge. Verizon and Equinix have publicly admitted that edge computing demand has stalled. We will continue to see pockets of deployments, but it will take a while for edge computing to boom without a new application to drive MEC needs.
Why 2023 isn’t the year for private 5G networks
Private 5G networks are one of those technologies that have received much airtime. Most operators have made announcements, and many smaller companies offer solutions. The FAQ page of “AWS Private 5G” says the service supports LTE core and LTE radio units operating in the Citizens Band Radio Service band. They say there will be an option to shift to 5G without timelines.
The reason most deployments today use 4G LTE or CBRS in the US is the release of the CBRS spectrum, which makes it easier for enterprises to utilize. The other reason is that improvements in Wi-Fi 6/6E/7 make it suitable for many applications where the 6GHz spectrum is available.
There is no point in comparing future versions of 5G with older versions of Wi-Fi. Both technologies will exist. There will be more private 5G deployments in the future, but it will be a slow roll, unlike most predictions for 2023.
Kubernetes can’t solve the problem yet
For 5G applications to thrive and benefit from lower latency and network determinism, one must deploy them as virtual or container network functions on the edge nodes. Kubernetes was thrown into the mix to solve this problem.
However, the edge is not just a closer cloud. Kubernetes struggles in resource-constrained edge environments, so along came efforts to shrink it, including MicroK8s, K3s, KubeEdge and other variants.
Hypervisors introduce performance overhead, defeating the purpose of bringing applications closer to the user. To counter this, we threw in DPDK, SR-IOV, PCI-passthrough and other solutions for processing packets at line rates. Latency-sensitive applications need immediate access to edge resources. To make that happen, we threw in CPU-pinning, HugePages and other techniques to provide applications with exclusive access to resources.
At the rate at which we are throwing solutions to solve the edge problems we have created, we might as well throw the kitchen sink in. All of this adds complexity, requiring intervention from highly skilled resources. We are bound to figure it out and wrap it in a package for 5G solutions to consume, but it will take some time for these solutions to mature in 2023.
Playing devil’s advocate on the edge
IoT, metaverse, video analytics, robotics, private 5G, gaming and other applications were all touted at some point as the killer app for 5G and edge computing. They might still be the much-needed impetus to get these network technologies into the commercial market.
We must experiment and plan to adopt these solutions on a large scale. In the meantime, ChatGPT seems to work just fine from our laptops without the need for 5G or edge computing.
Ritesh Mukherjee leads enterprise networking as senior vice president and general manager at Inseego. He is responsible for transforming networks and services with 5G, SDN/NFV, SASE, SD-WAN, cloud and IoT technologies.