As communications providers prepare to deliver high-speed connectivity to support new services and use cases, the demand for low-latency, high-bandwidth, scalable networks continues to grow. This may lead to software-defined infrastructure and cloud native approaches to meet 5G requirements. Here’s a look at what Linux Foundation (LF) is working on to create an open source ecosystem for the same.
Fig. 1: LF open source component projects for 5G
Successful open source developments depend on complete life-cycles of the projects that the market will adopt and deploy. There is an exciting convergence in the networking industry around open source, and the energy is palpable. In early 2021, the Linux Foundation (LF) announced a collaboration with Facebook to be managed under a neutral governance framework.
Magma, an open source software platform, enables operators to build and augment modern and efficient mobile networks at scale. It features an access-agnostic mobile packet core, advanced network automation and management tools, and the ability to integrate with existing LTE networks. It has use cases across both virtual and container network functions (xNFs) including fixed wireless access, carrier Wi-Fi, private LTE and 5G, network expansion, and mobile broadband.
In all these major areas, the LF hosts over 700 projects that are either in a technology area or a market area where innovation can speed up ongoing work in India. The numbers are just staggering these days. Most of the Fortune 500 companies use our quarter of a million developers, contributing code to almost 500 plus projects. And then if you look at just the stats through LFX insight, 31 million lines of code were added between 1.1 million poll requests and 10,000 repositories.
The open source journey has been quite significant in the past decade, starting with virtualisation, desegregation, and software defined networking to control plane databases. We have come a long way in the Kubernetes cloud native and are now moving to intent based.
While earlier focus was to develop projects in technology, building blocks for open source, in the next decade we are going to see all these open source projects working together for a common solution. The problem we are trying to solve this year is how to move code into production as quickly as possible.
While there are projects like Anuket, which is in LF networking, projects which are LFH can solve exactly this problem. They look at blueprints and how to comply with a spec, do open verification, submit the test results and do open interrupt, testing, training, and certification. So that’s kind of how we are seeing most of the evolution in open source this year.
Fig. 2: Focus shifting from building blocks to putting E2E OSS to use
LF networking has a unique perspective with the broadest set of projects that make up the diverse and evolving open source networking stack. It provides platforms and building blocks across the networking industry that enable rapid interoperability, deployment, and adoption, and is the nexus for 5G innovation and integration. It has now tapped into industry efforts to structure a new initiative to develop 5G Super Blueprints for the ecosystem. Major integrations between the building blocks are now underway between ONAP and ORAN, Akraino and Magma, Anuket and Kubernetes, and more.
The 5G super blueprint is what is going to integrate a set of projects all the way from access to core. ‘Super’ means that we are integrating multiple projects, umbrellas (such as LF Edge, Magma, CNCF, O-RAN Alliance, and LF Energy) with an end-to-end structure for the underlying facilities and application layers throughout edge, gain access to, and core.
This end-to-end combination allows leading market usage cases such as repaired wireless, mobile broadband, personal 5G, multi-access, IoT, voice services, and network slicing. In short, 5G Super Blueprints are a vehicle to team up and produce end-to-end 5G options.
With availability of customised solutions, there is a tremendous opportunity for the market. This is how the verticals are taking advantage of edge computing. Now healthcare and smart cities are towards the lower end of the spectrum. But it will need a little bit of regulations and process controls to update before adoption happens.
LF edge projects
There are two types of edges—user edge and service provider edge. A user age is dedicated and operated by the user. A service provider edge is shared as a service, but then not all usages are the same. You can either have a very constrained device edge, which could be microcontroller based embedded compute, or you could have a gateway type edge. You know in the user control or an on-prem data centre edge, which is secure in a factory building.
Fig. 3: LF edge projects
And what separates the two is the last mile networks. So, under the base station or in a smart central office, which is the access edge is not hard to know. Edge compute is for applications that need 20 milliseconds or less latency.
The edge project stack is a very important attribute. For example, if a sensor wakes up and dumps data to an Amazon cloud every week, it’s an IoT application. It’s not an edge application, and vice versa. Hence, the LF edge as an umbrella is unifying the frameworks.
Akraino R4 blueprints
Akraino is a blueprint concept. The open community comes together to define a use case. And then, through open governance, tests out the use case and interoperability all the way from hardware to the application, and then submits the results so that developers can replicate it.
That’s a blueprint, a declarative configuration, with CICD documentation. The one that has become very popular with telcos is PCEI (public cloud edge interface).
So, the question arises, how the public cloud and the service provider at the edge have a common set of interfaces? There are no standards that exist.
ORAN use cases include the slicing and the disaggregation that comes with the open RAM. And then on the Kubernetes, it’s the cloud, multi-cloud hybrid layer below with the standard compliance and then ONAP, which is the largest and the most important. It goes through 5G network slicing and all the use cases for telecom. This is really the fundamental platform, 10 plus million lines of code that is at the heart of network automation and network automation is a must for 5G.
So as a global community, ONAP has matured and is one of the most important projects to enable end to end systems.
In case of intermittent connectivity, the user needs an on-prem framework, which is Eve or edge virtualisation. EdgeXFoundry, the open source vendor neutral, is by far the largest framework for IoT. It abstracts southbound and northbound—all the messaging protocols and everything northbound—and this is where building automation, smart process control, water cities happens.
LF networking may now be working on about ten projects—Magma cloud native computing foundation, Kubernetes, or a software community. We are taking these communities and doing an open 5G super blueprint integration.
Fig. 4: The third phase of 5G modernisation
How it comes together
The 5G super blueprint comes pre-baked and only a small set of integration and support is needed by the community or their vendors. So, instead of multiple vendors doing things in different ways, now you have it done in the community and then the vendor can provide innovation and differentiation on top.
5G will modernise today’s energy grid with awareness monitoring across distribution systems.
It will roll out in three phases: 5G core+MEC phase, 5G E2E+MEC phase, and 5G+Slicing+O-RAN-SC +MEC phase.
In the first phase the communities are using the emulators on the RAN side, but Magma is being inserted with ONAP and Anuket on Kubernetes. There are some components from Intel called openness also coming in but, in general, those are the big new components.
The second phase is bringing in a commercial genome piece. The RAN side of things are integrated, and then you run a slice through it. So, the end-to-end network slicing comes in.
The last phase, which is towards the end of the year, is when ORAN comes in. What’s already happened is ONAP and ORAN are already deeply integrated with interfaces. There is a lot of component reuse between projects because these are all microservices, containerised modules, and you can just use them as needed and it will just get you a head start. And that’s already happening.
Completing the solution stacks to integrate a fully open networking and edge stack, speeding up ‘backend interior and deployment process’ with open compliance, integrated deployment in vertical markets to start embracing, utilising and building on networking, cloud, edge stacks are seen as the top priorities.
To conclude, blueprints are fairly matured and deployed. “Most of the projects are live in production. The first step is that it needs to happen. In several use cases, the 5G node is now merged from the open-ended interface side of things. And so, Magma itself, as an open mobile packet core in the Q3 timeframe, will be fairly mature.
The testing of these blueprints should be ready by the end of the year. Till then it is a proprietary mobile packet core. Pretty much everything that is at the higher layers of the stack is mature. At the lower layers, radio will stay proprietary, but the disaggregation concept is mature.
The article is based on the tech talk by Arpit Joshipura, GM, Networking Edge and IoT, Linux Foundation at Open Source Conference 2021