Edge computing and 5G is a new model of computing that will inevitably become the modus operandi for all organizations that rely on the cloud today. This is especially true of Telcos who will build and operate the new 5G networks of tomorrow.
What’s characteristic of edge applications is the low latency and high bandwidth they require. Devices at the edge have their processing, storage, and networking all configured on or near the device. These devices rely on an edge network for data processing. The edge removes the need to send a request to a centralized cloud and wait for a response. But to enable the potential of the edge, it will take a next generation of connectivity in the form of 5G.
In this post, we look at what’s new with 5G, the challenges associated with it, particularly around networking, and suggest ways to prepare and implement networking that is 5G-ready. To learn more, join the webinar Networking and Kubernetes in the World of 5G Edge which will be hosted by Platform9 on November 11 at 9 AM PDT.
4G to 5G – What changes?
The Edge requires a network with high bandwidth, low latency, and reliable performance. This is what 5G promises. While bandwidth gets a lot of attention, the most important of these factors is latency. Many edge devices such as self-driving cars need to make decisions in real-time or there are real-world consequences. These real-time decisions require data processing with extremely low latency.
To keep up with this pace, networking for the edge needs to function with equally low latency. A lift-and-shift approach to networking is not enough. The previous generation of cloud networking isn’t enough to satisfy the needs of the edge. What’s needed is to glean the best principles of cloud-native networking and implement a new kind of networking that is purpose-built for the edge.
OpenStack to K8s
Telcos are heavily invested in OpenStack and VMware. This is to avoid vendor lock-in and because of the need for virtualizing network functions. Today, the world has mass migrated to Kubernetes as the operating system of the enterprise. It is built on open source tools, and rather than virtualization, encourages containerization.
Telcos that may feel like just yesterday they migrated from physical servers and networking to Virtual Network Functions (VNFs) now find themselves on the cusp of another shift to Container Network Functions (CNFs). However, this shift need not be hasty and need not throw away years of constant improvement and learnings. Telcos can still find ways to run OpenStack alongside Kubernetes, and run virtualized network functions alongside CNFs.
Networking for the edge needs to be built on containers and on open source software. This doesn’t stop with Kubernetes, but extends to the ecosystem of tools that support the platform.
KubeVirt bridges the gap
One interesting open source solution that is very relevant to this conversation is KubeVirt. It enables organizations to manage VMs using Kubernetes, and to manage these VMs using the same Kubernetes commands used to manage containers. These are the kind of solutions that Telcos need as they look to create a new future while still extending past investments in technology.
The defining characteristic of next generation networking in a Kubernetes stack is the service mesh architecture. While the same principles of automated load balancing, service discovery, and separation of control and data planes are required at the edge, there are new challenges that need to be solved. The edge handles sensitive information about users, and this information needs to be processed at or near the edge itself. This requires additional protocols for security and isolation so that the hardware, firmware, and networking are tamper-proof.
The Edge needs to enable communication over a local network. Here again, as data is in transit, it needs to be encrypted. This all needs to happen while maintaining the real-time low latency. This is quite a feat.
Service mesh solutions are built for the cloud-native era and have the underpinnings to support the edge, but they need to be implemented differently for edge. The service mesh ecosystem is maturing, and will get there soon.
5G is now
5G and edge computing is no longer a futuristic technology trend to wait for. It is happening now, and it is being created by the innovators and leaders of today. The largest telcos are partnering with leading cloud-native and edge management platforms to bring their solutions to market sooner than their competitors.
During these defining times, it pays to be ready with a partner for edge networking. One that has worked with Telcos since the VMware and OpenStack days, and gets their unique challenges. But importantly, a partner that gets cloud-native computing and how its principles can be effectively applied to the new world of edge computing and 5G.
To learn more, join the webinar Networking and Kubernetes in the World of 5G Edge which will be hosted by Platform9 on November 11 at 9 AM PDT.