Categories
Uncategorized

Chaotic Intersection of 5G, Edge, and Cloud

New architecture paradigms always bring new challenges for adopters. Edge computing and 5G are no different. Carriers, Network Equipment Providers, Enterprises, ISVs, Hyperscalers, startups are all eyeing new market opportunities enabled by 5G and Edge computing. But there are several barriers and inter-dependencies across ecosystem players that are slowing the adoption. 5G Open Innovation Lab was created with a mission to identify new use cases, remove market and technology barriers and accelerate adoption of 5G and Edge computing.

The Problem

Telecom ecosystem players hold a network centric view of the edge where they provide 5G network infrastructure and services, leaving applications to others. Hyperscalers are building and leveraging their own edge locations and partnering with operators as needed to leverage telco central offices. Each hyperscaler has its own edge computing model and strategy pursuit, separate from others for competitive reasons. Multiple 5G consortiums have been spun up creating opportunistic alliances among the operators across the world. In addition, the recent opening of 3.5GHz spectrum from CBRS now allows enterprises to set up their own private 5G networks.

Where does this leave the enterprises, startups, and developers?

Enterprises, as developers and end customers, are forced to deal with the complexity of a fragmented ecosystem and a myriad of options. Investment in 5G and Edge computing applications is risky unless they can ensure future proofing those applications by abstracting them from the underlying infrastructure.

Developers want to build applications that are not tied to infrastructure and can work across all networks regardless of who owns or operates the network or whether it is public or private.

Startups are trying to address many of the gaps but have the arduous task of breaking into a legacy industry that has not quite figured out how to embrace modern technologies and development practices.

Cloud native technologies such as Kubernetes have seen massive adoption in the last few years. The core idea of Kubernetes was to have developers build apps, maintain apps, and deploy them in a scalable and reliable way without dealing with underlying infrastructure complexities. There have been attempts to use cloud native technologies to build 5G network infrastructure. Rakuten Mobile is worth mentioning as they are the first operator to open up the 5G network built using cloud native stack.

A hyperscaler like AWS can own and manage all of the infrastructure, facilities, supply chain, business model and provide cloud services with SLA. However, it is much harder to provide services with SLA on the edge. Imagine the complexity of the provider ecosystem on the edge (Telco’s, Cable providers, co-los, Enterprise DCs, hyperscaler owned PoPs etc), changing network conditions, varying footprint and infrastructure power constraints, and other logistical complexities and you’ll see building an “edge cloud” is an order of magnitude harder than the “cloud” as we know it. Therefore, hyperscalers are using the proven cloud model, striking partnerships with Telco’s and Cable companies to put their own gear in the edge locations and connect it back to cloud.

From a software perspective, what worked well in the mainstream computing world, starts to break down as we start to move to the edge. Microservices, Containers, Kubernetes, Serverless and others are seeing rapid adoption in the enterprise. They do not work the same way when you are dealing with new application requirements on the edge. A new framework is needed.

The Solution

Innovation demands a uniform, vendor-neutral way to build, deploy, and manage applications that are tiered across edge and cloud.

There have been multiple efforts in the industry to solve this problem. Many of these efforts are centered around extending the success of Kubernetes in the cloud to the edge as well. It’s worth mentioning cloud vendor agnostic efforts such as Rancher K3SKubeEdge and cloud vendor specific projects that tie proprietary device management (such as AWS GreenGrass, Azure IOT) to application deployment using standard K8S.

Here are some of the tenets of the solution that we believe is right for the 5G/Edge community:

1.      Unlike “cloud” owned by a few hyperscalers, the Edge would be a single federated cloud where developers will build and deploy once to a federated control plane. Hyperscalers, Telco, Cable providers need to be good citizens of the federated world along with other providers.

2.      Compute would be network driven. In other words, the network event will drive the “on-demand” application deployment (like event driven “Functions-as-a-service” in the cloud world)

3.      New challenges in security would need be addressed by the providers to build the trust in the edge ecosystem. The 1:1 trust model in the cloud world won’t work, requiring a federated trust approach.

4.      Edge deployments need a new software framework for application deployment to balance infrastructure capacity, latency, and cost/monetization model. For instance, providers may have “spot-pricing” of resources depending on available infrastructure capacity. Applications should be able to describe the resource needs and cost constraints to the framework, and the framework should take care of the rest.

We don’t have a set of standards to make “write once, deploy on any edge or any cloud” come true although CNCF driven projects are getting wide adoption across the cloud vendors. Given the complexity of the edge, unique application requirements, competition among operators, cable providers, and hyperscalers, it may be a long time before standards emerge covering edge and cloud.

A new set of players in the ecosystem may emerge and address the lack of standards by mediating across providers while providing a uniform interface to the developers.

At 5G Open Innovation Lab, we nurture innovators so together we can deliver a new world of 5G/Edge enabled applications to the world. By building a truly open ecosystem of partners – from Telco to Cloud and Edge Computing leaders – we hope to accelerate the adoption of this federated approach to 5G application development.