Has 2022 really been the year of virtualisation so far?
Firstly, what are NFVs? A standard definition would be separating out network functions (e.g. mobile core, routing, etc.) from the dedicated hardware appliances that have delivered them in the past, and redesigning them as software running over commercial off-the-shelf hardware.
Every telco has a cloud strategy, and NFV represents a critical element within that broader transformation. There’s an overlap between cloud and virtualisation. The virtualisation market as a whole is estimated to be in +30%/year growth for the next five years, with an approximate value of US$70-120bn market size by 2025.
It is interesting to see how virtualisation is now touching every part of the telecoms network, and wider ecosystem, when we look back at the the ETSI NFV paper launched in 2012.
Now, the UK Department for Digital, Culture, Media and Sport (DCMS) has invested in the FRANC competition – to incentivise industry to create new products and services to unlock the full potential of Open RAN – with the rationale being that virtualisation and Open RAN are a collaborative way to open up the network equipment and software landscape, stimulating greater competition and innovation. As the technology evolves and automation emerges, this approach offers more and more promise.
So, what does the future hold in store for virtualisation? Should we be full of optimism or braced for realism?
Here, three CW Journal authors share their thoughts on the topic:
Paul Crane, former BT Networks Research Director, and Visiting Professor at the University of Surrey.
If we project forward to 2030, we can think about a telecoms network as being effectively a set of distributed compute resource, where we put network functions or application functions at the right point in the network at the right time to deliver the service that we need. Concepts like edge are a transition towards that vision.
BT have already implemented their cloud native core network. They are currently migrating applications onto that core network – for instance, broadband authentication, packet gateways, CDN, but the big application is the mobile core network, and that should be completed in Q3 2022.
That is the biggest and the most challenging thing BT are doing at the moment with virtualisation, particularly when you take into account the resilience and reliability metrics that we need to support.
The other area that we’re looking at is the radio access network and Open RAN. Why the fuss? Lots of announcements in the press in the past year – from Ericsson’s white paper on security concerns, Rakuten prepping the symphony input, NEC setting up R&D facilities in the UK. The main reasons given for the excitement are cost saving and innovation.
Cost saving is a really interesting metric to use as I have not seen any real evidence to suggest that, for an operator like BT, that isn’t a green field. BT are running 2G, 3G, 4G and 5G networks – it’s not obvious that retrofitting an Open RAN approach will result in any cost savings compared to a traditional commercial approach. Evidence hasn’t been seen yet, but I suspect that as the ecosystem matures the economic case will become clear.
Turning to Open RAN standards, a key issue is that we need a standardisation that process is going to result in, and maintain, openness. The standards aren’t yet finished, and BT is activity contributing to key aspects to drive openness.
So, should we be implementing Open RAN?
The question for BT is when – when is it most appropriate, when will it improve customer experience?
For BT to change to an open architecture in its main network, they’re going to need something that’s going to prove to be resilient, prove to be reliable, is going to have benefit – commercial benefit – of beyond a standard approach, and provide a route for future innovation.
In terms of future innovation, the thing that is exciting me the most is the RIC – the RAN Intelligence Controller – because this is where BT can potentially differentiate and improve customer experience.
For new services, it could be simple things like very accurate device location on the RAN, or it could be being able to plan your network in a much more intelligent way based on near real-time information from the network.
BT are enthusiastic about Open RAN and do think it’s the future, but for the main core networks they’re saying not yet.
Open RAN has emerged as the top topic in telecoms, and has even made on the agenda of the EU, UK and USA governments. We’ve even seen politicians on the TV talking intelligently about telecoms architectures. They’re interested because of supply chain diversification. Mobile is fundamental to the digital economy and politicians are concerned about the resilience of the supply chain. For a radio access network, European operators have two, maybe three, choices of suppliers. The EU is therefore investing research funding into Open RAN and likewise, the US has a bill going through the house for $1.5bn for funding Open RAN-type equipment.
One of the first things announced by the UK Government was the FRANC competition, the winners of which was announced in December. There is a target of 35% of traffic to be run on Open RAN architectures by 2033. This has heightened interest in Open RAN inside the industry.
What is BT doing?
BT has a long history in virtualisation from driving the initial concepts through to the practical implementation of the technology. At BTs R&D headquarters at Adastral Park, there is Telco Infra Project (TIP) Community Lab. Also at Adastral Park, BT has deployed an Open RAN small cells network for infrastructure innovation projects and 5G private network for service and application co-innovation.
The applied research team are also exploring where else virtualisation can benefit customer experience and overcome emerging issues.
My experience so far:
So, if you take a cloud core network, what we’re finding is that BT built ours and other operators have done the same, but all those networks, all that infrastructure, is slightly different and bespoke to an individual operator.
Now the question is, is that a sustainable approach? Is there a risk that an operator ends up locked into a set of vendors just by a different method?
I’ve not seen any evidence so far that virtualisation has reduced our energy usage. In fact, the evidence is pointing in the other direction now, and that’s a key thing for any operator. Energy consumption is emerging as at top concern, both from a cost and an environmental point of view. This is a key area for research and innovation.
Network security is going to become an increasingly important issue. Now telecommunications networks are using standard IT equipment they are much more exposed. Ensuring security and resilience is another key topic for the industry.
Finally, one of the challenges discovered when building the cloud platform is the skills gap. There are very few people that understand telecoms and understand IT. This is another key issue the industry needs to tackle.
Paul Rhodes, OpenRAN and 5G Lead, World Wide Technology
Over the past decade WWT have done a lot of work on virtualisation and containerisation, and over these years we have really taken this from the core to the edge, going through a number of technology changes.
The Advanced Technology Centre (ATC) is our secret weapon – it’s a centre of innovation, between partners and customers with £500m invested over the years and hundreds of original equipment manufacturers (OEM). We mainly offer on an on-demand basis; a service to key customers to use the lab to build remotely.
Is virtualisation at its peak?
Most of what we’re deploying today is virtualised, so it’s definitely ready for prime time. We’re more looking at things like manual intervention vs zero touch, so how do we deploy this virtualised system as quickly as possible. As we look at the core or Open RAN deployment, if you take a step back you clearly understand what that looks like, but in the day-to-day it is not always clear. What we do at WWT is to facilitate all of this.
Cross core RAN and service layer are pretty complex environments – – a plethora of different vendors on the infrastructure and software sides. So, over the years you’ve created a lot of silos for the different vendors. One vendor for the core, one vendor for the DMS, another for the RAN deployment. They might all have a different approach on architecture deployment and their own software.
A lot of our focus time is spent on simplifying the architectures – both the idea to certify different vendors using the same infrastructure stack and a set of consolidated orchestration tooling. This means that whether software is deployed for the core or the RAN, the same specialist team and tooling can be used.
All services have basically the same requirements – they need a VIM, a Management Layer, some compute, some storage and networking.
Merging these things horizontally can reduce cost, and promote agility. Your team can buy in thousands of servers at a time rather than tens. There is commonality in the hardware and software.
By starting with a simple blueprint approach, and then shipping out at scale, you can reduce delivery time by 70% through a logical and stepwise process. This enables more agility in delivery.
With our UK customers, now the infrastructure is in they are working on automating the VNS, VCU and VDU so that they can see how that looks.
Continuous integration/continuous development has not really been fully standardised. A common approach is needed. Continuous test needs to be added to deliver CI/CD/CT, particularly with Open RAN.
Telco operators are conservative in their approach of how to take software into their network, because they want robustness. There is a methodical process of monthly/quarterly/annual/ emergency patches. But the telco space is now moving to “Here’s a patch, take it”, so you’ve got to work out how to do the ‘best of breed’ approach that merges the culture of the telco and the potentials of the technologies.
You’re taking a relatively young and enthusiastic cloud approach, and a more mature approach in terms of telco, and it’s very hard for us to embrace change, but absolutely that’s where we’re going now. A great system integrator can seamlessly talk both the telco customers’ and IT vendors language and marry the two harmoniously.
One of the main changes from last year, and that is really accelerating in 2022, is that virtualisation vendors are bridging the gap between general infrastructure management and orchestration. So telco operators now have much tighter control when it comes to orchestration workloads on top of a telco cloud or RAN cloud. The other thing that we’ve seen is release vendors making acquisitions in the RIC space, building the utilisation layer.
There are big cost savings from time to market, but also resource saving from a people time perspective when tuning the service and automation.
The last few years has seen the implementation of kubernetes in the cloud, enabling great savings.
It’s all about the ability to launch that workload wherever you want it to be. That’s what will give us the value output – if it’s all out on one location, right at the edge, it’s a private network. You put it all in a centralised location, it’s just a traditional monolithic network. But now you can fully cloudify that and run it on AWS if that’s what you want.
David Martin, experienced telecoms analyst (STL partners)
At the previous update to the Telco Cloud Deployment Tracker in November 2021, this contained information on 950 live, commercial and public-domain deployments of telco-cloud, predominantly in Tier 1 telcos.
The full extent of telco network virtualisation is even greater than shown in the Tracker: as it contains only public domain information, it represents the tip of the iceberg – although it is a useful cross-section of all that is going on.
Virtualisation is not new and could, at first sight, even appear to be in decline. Total new telco cloud deployments peaked in 2019, driven by mobile core and SD-WAN. Core has remained the leading category from 2020 onwards, linked to 5G network launches based on the Non-standalone (NSA) 5G core.
The ‘decline’ reflects the fact that big foundational projects around SDN and NFV MANO (Management and Orchestration) have been completed.
However, cloud-native network functions (CNFs) will drive a new growth phase. CNFs represent a second phase of virtualisation, with network functions built for the cloud from the ground up and based on containerised software micro-services, with open interfaces allowing for multi-vendor operations and more programmability.
The main CNFs that will drive this second growth phase will be the 5G Standalone (SA) core (SA 5GC) and virtualised/Open RAN. We expect the number of SA 5GC deployments by major telcos in 2022 to exceed those of 5G NSA, which drove the first wave of 5G network launches in 2019 and 2020.
The number of cloud-native, virtualised RAN deployments has been modest so far. But they include some large-scale networks, such as: Rakuten Mobile in Japan; and Vodafone Idea and Bharti Airtel in India. These are examples of open RAN, built with open interfaces designed to facilitate multi-vendor deployments. Examples of single-vendor virtualised RAN include Verizon in the US, with both Ericsson and Samsung.
In historical perspective, the US and China are the countries with the largest numbers of telco cloud deployments. Vodafone is the most active operator in the VNF/CNF space, followed by China Mobile.
In conclusion, it is not strictly accurate to describe 2022 as the ‘year of virtualisation’ as this has been going on for ten years or more. However, 2022 will see a ramping up of the second phase of virtualisation: the telco cloud built on cloud-native software principles.
The major driver of this is 5G: both the SA core, and open and virtualised RAN.v
David manages the Telco Cloud research stream at STL Partners, writing recent reports on topics including the 5G core, open RAN and network disaggregation. He also manages STL's 'Telco Cloud Tracker' product: a continuously updated database and interactive analytical tool looking at live deployments of telco cloud and Virtualised and Cloud-Native Network Functions (VNFs / CNFs) at major telco networks worldwide. He is a telecoms analyst of 23 years’ experience and, in addition to STL Partners, has worked for Ovum (now Omdia) and Analysys Mason. David gained a First Class degree in French and German at Trinity Hall, Cambridge, and then spent nearly five years at the same college doing PhD research on 19th century Romantic, Gérard de Nerval.
Paul Rhodes is a senior technical sales leader with 28 years’ experience spanning strategy, radio and fixed networks, building small cell networks, sales and negotiating contracts. Paul leads WWT’s 5G OpenRAN EMEA engagement. Previous to this role, Paul held senior commercial and technology positions with Ericsson, Samsung and Nokia, working with all the UK MNOs, European Operator groups and Canadian Carriers. Paul has also consulted on FWA and Fixed-Mobile convergence, and has delivered advanced mobile solutions across Europe, North America and Asia.