Kubernetes has gained popularity as the point of intersection for all things cloud-native. As we move to a world of 5G, Kubernetes has a vital role to play in enabling tomorrow’s 5G networks. Having established itself as the de-facto option to manage centralized cloud datacenters, Kubernetes now has some distance to travel as it finds usage in the middle and outer edges of the network. Let’s look at Kubernetes’ role in enabling tomorrow’s 5G networks.
Edge devices are a balancing act
Edge computing is about taking processing away from the centralized datacenter and moving it to the outer devices. The goal of this activity is to enable real-time processing to happen on the device without having to send data back and forth across the network. While 5G networks will quicken data speed and reduce latency, the biggest improvement in real-time processing will be due to moving processing to edge devices.
The edge is powered by small-scale devices with low hardware specs that are just enough to perform the required operations. It is a fine balance. Too much power and the edge devices would not be sustainable in terms of energy requirements and costs. Remember, edge devices are operated by the thousands across numerous different locations. They need to be kept nimble and agile.
On the other hand, edge devices need sufficient power to be able to process jobs on their own without help from the centralized servers. It also needs the right memory capacity to store data temporarily for its processing needs and periodically send data for storage in a central server. It needs airtight security controls over the network so it can communicate with the centralized server and any middle-layer devices and networks.
These devices are capable of hosting applications, an operating system, containers, and, importantly, a container orchestration tool like Kubernetes. But it’s not that the edge should be run on Kubernetes just because it is possible. Rather, the edge makes Kubernetes a necessity.
Kubernetes, and Kubernetes alone, for the 5G edge
Kubernetes is the only solution that is secure enough, robust enough, and mature enough to cope with the complexity of the edge. Kubernetes has now spent a good five years on the battlefield, getting tested in every possible scenario. It has inspired an ecosystem of tools around it for various purposes. There is Prometheus for monitoring, Istio and Linkerd for a service mesh, Fluentd for logging, Jaeger for distributed tracing, Helm for package management, and the list is too long to cram in here. It’s safe to say that Kubernetes has shown in the cloud how edge should be done.
Attempting to rebuild edge networks from scratch is a recipe for disaster. With the numerous cloud vendor organizations now part of the CNCF, it’s no surprise Kubernetes enjoys the kind of industry-wide favor it does today. This kind of “across the aisle” cooperation is required at the edge as well. It calls for a level of openness that is unnatural to telcos. Thus far, telcos have focused exclusively on proprietary technologies, closed innovation, and cutthroat competition. In the new world of edge computing, they will need to rub shoulders with each other like the cloud vendors of today do — sharing seats on an oversight committee and working together on open source projects. All this calls for a radical openness to innovation. There will still be room for differentiating their services against the competition’s, but that’s a much smaller focus, especially at the start.
The middle layer of the edge
Apart from the devices at the very edge of the network, there is a crucial middle layer of devices that are primarily used to aggregate and transmit data from the edge to the datacenter. They act as a bridge between the two ends of the network. These devices are the “fiber in the ground,” as Anna Claibourne of PacketFabric describes. Traditionally, they have been inflexible as they were based on physical equipment. However, that is about to drastically change in the world of 5G.
VNFs and CNFs for dynamic networks
Over the past decade, telcos have spent time transitioning from purely physical networking to virtual networks in the form of virtual network functions (VNFs). This brought some flexibility and speed of operations, but this is still not enough for the pace of 5G operations. 5G will require dynamic network provisioning and management. For example, users would come to expect on-demand bandwidth from 5G. This means saving costs with low bandwidth 5G for activities like web browsing but paying a premium for data-intensive activities like mobile gaming. This is called network slicing, and it is the next wave of innovation in mobile data offerings. Telcos are gearing up to provide this flexibility to their users. However, they still lack the infrastructure to make this possible.
With 5G, networking will move from VNFs to CNFs (container network functions) that are not run on VMs but on container instances. Unlike VMs that are long-running instances, containers are meant to be ephemeral. They support services that can be as short as a few seconds end-to-end. They take fractions of a second to startup and to be destroyed. They don’t incur additional costs for each action of creating or destroying instances. They are nimble and capable of supporting the dynamic networking needs of tomorrow’s 5G.
Here again, Kubernetes is the management layer for these CNFs. But the big surprise here is that Kubernetes is not just used to manage CNFs but VNFs as well. KubeVirt is an open-source project that does this. It allows VMs to be run alongside containers and to be managed using Kubernetes.
Preserving past investments
This is crucial for telcos as they have invested a lot of time and resources into setting up and maintaining VNFs over the past decade. Most of this is run either in OpenStack or VMware. Now that Kubernetes has taken over the reins as the preferred operating system for the edge, telcos want to sustain their investments in these older technologies. Not just that, for some organizations, compliance and regulatory reasons and technical debt mean that they need to continue to operate VNFs for the foreseeable future. Thankfully, Kubernetes provides the common ground between the old and the new, and in doing so, eases the transition to 5G.
Managing the management plane
Though Kubernetes is the operating system across the three levels of edge computing (datacenter, middle, and last-mile), an underlying problem that still needs to be solved is about managing and monitoring Kubernetes itself. If the same solution is replicated and used across layers, it also makes sense to have a common management plane for the solution. This not only reduces effort but it brings much-needed consistency to a complex system.
This underlying management layer for Kubernetes is still in the making. Many cloud vendors big and small are vying to become the best place for telcos looking to deploy and manage Kubernetes for 5G edge networks. Without a doubt, the big three cloud vendors — AWS, Azure, and Google Cloud — are reaching out to telcos with dedicated marketing and sales pitches. However, looking beyond them, there are numerous smaller vendors that are building innovative solutions. These include Platform9, Weaveworks, Mirantis, and more.
5G powered by Kubernetes is the future
It is an exciting time to be in cloud computing or telecommunications. Things are changing very fast, and there is a whole world of opportunities waiting to be had. The winners will be those who can combine the best of these disparate technologies and build a new kind of network — 5G powered by Kubernetes.
Featured image: Pixabay