Telco Cloud Deployment Tracker: Deploying NFs on public cloud without losing control

In this update, we present a review of telco cloud deployments for the whole of 2022 and discuss trends that will shape the year ahead. Fewer deployments than expected were completed in 2022. The main reason for this was a delay in previously announced 5G Standalone (SA) core roll-outs, for reasons we have analysed in a previous report. However, we expect these deployments to be largely completed in 2023. 

We also review deployments of NFs on the public cloud in 2022. While few in number, they are significant in scope, and illustrate ways in which telcos of different types can deploy NFs on public cloud while retaining control over the management and ongoing development of those NFs.

Enter your details below to download an extract of the report

CNFs on the public cloud: Recent deployments illustrate how to avoid hyperscaler lock-in

Few telcos have yet deployed critical network functions on the hyperscale cloud, as discussed in this report. However, significant new deployments did go live in 2022, as did tests and pilots, involving all three hyperscalers:​

Recent deployments and trials of CNFs on public cloud

Source: STL Partners

In our recently published Telco Cloud Manifesto 2.0, we argued that telcos thinking of outsourcing telco cloud (i.e. both VNFs/CNFs and cloud infrastructure) to hyperscalers should not do so as a simple alternative to evolving their own software development skills and cloud operational processes. In order to avoid a potentially crippling dependency on their hyperscaler partners, it is essential for operators to maintain control over the development and orchestration of their critical NFs and cloud infrastructure while delivering services across a combination of the private cloud and potentially multiple public clouds. In contrast to a simple outsourcing model, the deployments on public cloud in 2022 reflect different modes of exploiting the resources and potential of the cloud while maintaining control over NF development and potential MEC use cases. The telcos involved retain control because only specific parts of the cloud stack are handed over to the hyperscale platform; and, within that, the telcos also retain control over variable elements such as orchestration, NF development, physical infrastructure or the virtualisation layer.

In this report, we discuss the models which the telcos above have followed to migrate their network workloads onto the public cloud and how this move fits their overall virtualisation strategies.

Previous telco cloud tracker releases and related research

Enter your details below to download an extract of the report

Telco Cloud Deployment Tracker: Will vRAN eclipse pure open RAN?

Is vRAN good enough for now?

In this October 2022 update to STL Partners’ Telco Cloud Deployment Tracker, we present data and analysis on progress with deployments of vRAN and open RAN. It is fair to say that open RAN (virtualised AND disaggregated RAN) deployments have not happened at the pace that STL Partners and many others had forecast. In parallel, some very significant deployments and developments are occurring with vRAN (virtualised NOT disaggregated RAN). Is open RAN a networking ideal that is not yet, or never will be, deployed in its purest form?

In our Telco Cloud Deployment Tracker, we track deployments of three types of virtualised RAN:

  1. Open RAN / O-RAN: Open, disaggregated, virtualised / cloud-native, with baseband (BU) functions distributed between a Central Unit (CU: control plane functions) and Distributed Unit (DU: data plane functions)
  2. vRAN: Virtualised and distributed CU/DU, with open interfaces but implemented as an integrated, single-vendor platform
  3. Cloud RAN (C-RAN): Single-vendor, virtualised / centralised BU, or CU only, with proprietary / closed interfaces

Cloud RAN is the most limited form of virtualised RAN: it is based on porting part or all of the functionality of the legacy, appliance-based BU into a Virtual Machine (VM). vRAN and open RAN are much more significant, in both technology and business-model terms, breaking open all parts of the RAN to more competition and opportunities for innovation. They are also cloud-native functions (CNFs) rather than VM-based.

Enter your details below to request an extract of the report

2022 was meant to be the breakthrough year for open RAN: what happened?

  • Of the eight deployments of open RAN we were expecting to go live in 2022 (shown in the chart below), only three had done so by the time of writing.
  • Two of these were on the same network: Altiostar and Mavenir RAN platforms at DISH. The other was a converged Parallel Wireless 2G / 3G RAN deployment for Orange Central African Republic.
  • This is hardly the wave of 5G open RAN, macro-network roll-outs that the likes of Deutsche Telekom, Orange, Telefónica and Vodafone originally committed to for 2022. What has gone wrong?
  • Open RAN has come up against a number of thorny technological and operational challenges, which are well known to open RAN watchers:
    • integration challenges and costs
    • hardware performance and optimisation
    • immature ecosystem and unclear lines of accountability when things go wrong
    • unproven at scale, and absence of economies of scale
    • energy efficiency shortcomings
    • need to transform the operating model and processes
    • pressured 5G deployment and Huawei replacement timelines
    • absence of mature, open, horizontal telco cloud platforms supporting CNFs.
  • Over and above these factors, open RAN is arguably not essential for most of the 5G use cases it was expected to support.
  • This can be gauged by looking at some of the many open RAN trials that have not yet resulted in commercial deployments.

Global deployments of C-RAN, vRAN and open RAN, 2016 to 2023

Image shows global deployments of C-RAN, vRAN and open RAN, 2016 to 2023

Source: STL Partners

Previous telco cloud tracker releases and related research

Enter your details below to request an extract of the report

O-RAN: What is it worth?

Introducing STL Partners’ O-RAN Market Forecast

This capex forecast is STL Partners’ first attempt at estimating the value of the O-RAN market.

  • This is STL Partners’ first O-RAN market value forecast
  • It is based on analysis of telco RAN capex and projected investment pathways for O-RAN
  • The assumptions are informed by public announcements, private discussions and the opinions of our Telco Cloud team
  • We look forward to developing it further based on client feedback

Enter your details below to request an extract of the report

What is O-RAN?

We define O-RAN as virtualised, disaggregated, open-interface architectures.

  • Our O-RAN capex forecasts cover virtualised, disaggregated, open-interface architectures in the Radio Access Network
  • They do not include vRAN or O-RAN compliant but single vendor deployments

O-RAN definition open RAN

O-RAN will account for 76% of active RAN capex by 2030

As mobile operators upgrade their 4G networks and invest in new 5G infrastructure, they can continue purchasing single vendor legacy RAN equipment or opt for multi-vendor open-standard O-RAN solutions.

Each telco will determine its O-RAN roadmap based on its specific circumstances (footprint, network evolution, rural coverage, regulatory pressure, etc)1. For the purpose of this top-level O-RAN capex forecast, STL has defined four broad pathways for transitioning from legacy RAN/vRAN to O-RAN and categorised each of the top 40 mobile operators in one of the pathways, based on their announced or suspected O-RAN strategy.

Through telcos’ projected mobile capex and the pathway categorisation, we estimate that by 2026 annual sales of O-RAN active network elements (including equipment and software) will reach USD12 billion, or 21% of all active RAN capex (excluding passive infrastructure). By 2030, these will reach USD43 billion and 76%, respectively.

Total annual O-RAN capex spend

Table of content

  • Executive summary
    • O-RAN forecast 2020-2030
    • Brownfield vs greenfield
    • Four migration pathways
  • Modelling assumptions
  • Migration pathways
    • Committed O-RAN-philes
    • NEP-otists
    • Leap-froggers
    • Industrial O-RAN
  • Next steps

 

Enter your details below to request an extract of the report

Open RAN: What should telcos do?

————————————————————————————————————–

Related webinar: Open RAN: What should telcos do?

In this webinar STL Partners addressed the three most important sub-components of Open RAN (open-RAN, vRAN and C-RAN) and how they interact to enable a new, virtualized, less vendor-dominated RAN ecosystem. The webinar covered:

* Why Open RAN matters – and why it will be about 4G (not 5G) in the short term
* Data-led overview of existing Open RAN initiatives and challenges
* Our recommended deployment strategies for operators
* What the vendors are up to – and how we expect that to change

Date: Tuesday 4th August 2020
Time: 4pm GMT

Access the video recording and presentation slides

————————————————————————————————————————————————————————-

For the report chart pack download the additional file on the left

What is the open RAN and why does it matter?

The open RAN’ encompasses a group of technological approaches that are designed to make the radio access network (RAN) more cost effective and flexible. It involves a shift away from traditional, proprietary radio hardware and network architectures, driven by single vendors, towards new, virtualised platforms and a more open vendor ecosystem.

Legacy RAN: single-vendor and inflexible

The traditional, legacy radio access network (RAN) uses dedicated hardware to deliver the baseband function (modulation and management of the frequency range used for cellular network transmission), along with proprietary interfaces (typically based on the Common Public Radio Interface (CPRI) standard) for the fronthaul from the baseband unit (BBU) to the remote radio unit (RRU) at the top of the transmitter mast.

Figure 1: Legacy RAN architecture

Source: STL Partners

This means that, typically, telcos have needed to buy the baseband and the radio from a single vendor, with the market presently dominated largely by the ‘big three’ (Ericsson, Huawei and Nokia), together with a smaller market share for Samsung and ZTE.

The architecture of the legacy RAN – with BBUs typically but not always at every cell site – has many limitations:

  • It is resource-intensive and energy-inefficient – employing a mass of redundant equipment operating at well below capacity most of the time, while consuming a lot of power
  • It is expensive, as telcos are obliged to purchase and operate a large inventory of physical kit from a limited number of suppliers, which keeps the prices high
  • It is inflexible, as telcos are unable to deploy to new and varied sites – e.g. macro-cells, small cells and micro-cells with different radios and frequency ranges – in an agile and cost-effective manner
  • It is more costly to manage and maintain, as there is less automation and more physical kit to support, requiring personnel to be sent out to remote sites
  • It is not very programmable to support the varied frequency, latency and bandwidth demands of different services.

Enter your details below to request an extract of the report

Moving to the open RAN: C-RAN, vRAN and open-RAN

There are now many distinct technologies and standards emerging in the radio access space that involve a shift away from traditional, proprietary radio hardware and network architectures, driven by single vendors, towards new, virtualised platforms and a more open vendor ecosystem.

We have adopted ‘the open RAN’ as an umbrella term which encompasses all of these technologies. Together, they are expected to make the RAN more cost effective and flexible. The three most important sub-components of the open RAN are C-RAN, vRAN and open-RAN.

Centralised RAN (C-RAN), also known as cloud RAN, involves distributing and centralising the baseband functionality across different telco edge, aggregation and core locations, and in the telco cloud, so that baseband processing for multiple sites can be carried out in different locations, nearer or further to the end user.

This enables more effective control and programming of capacity, latency, spectrum usage and service quality, including in support of 5G core-enabled technologies and services such as network slicing, URLLC, etc. In particular, baseband functionality can be split between more centralised sites (central baseband units – CU) and more distributed sites (distributed unit – DU) in much the same way, and for a similar purpose, as the split between centralised control planes and distributed user planes in the mobile core, as illustrated below:

Figure 2: Centralised RAN (C-RAN) architecture

Cloud RAN architecture

Source: STL Partners

Virtual RAN (vRAN) involves virtualising (and now also containerising) the BBU so that it is run as software on generic hardware (General Purpose Processing – GPP) platforms. This enables the baseband software and hardware, and even different components of them, to be supplied by different vendors.

Figure 3: Virtual RAN (vRAN) architecture

vRAN architecture

Source: STL Partners

Open-RANnote the hyphenation – involves replacing the vendor-proprietary interfaces between the BBU and the RRU with open standards. This enables BBUs (and parts thereof) from one or multiple vendors to interoperate with radios from other vendors, resulting in a fully disaggregated RAN:

Figure 4: Open-RAN architecture

Open-RAN architecture

Source: STL Partners

 

RAN terminology: clearing up confusion

You will have noticed that the technologies above have similar-sounding names and overlapping definitions. To add to potential confusion, they are often deployed together.

Figure 5: The open RAN Venn – How C-RAN, vRAN and open-RAN fit together

Open-RAN venn: open-RAN inside vRAN inside C-RAN

Source: STL Partners

As the above diagram illustrates, all forms of the open RAN involve C-RAN, but only a subset of C-RAN involves virtualisation of the baseband function (vRAN); and only a subset of vRAN involves disaggregation of the BBU and RRU (open-RAN).

To help eliminate ambiguity we are adopting the typographical convention ‘open-RAN’ to convey the narrower meaning: disaggregation of the BBU and RRU facilitated by open interfaces. Similarly, where we are dealing with deployments or architectures that involve vRAN and / or cloud RAN but not open-RAN in the narrower sense, we refer to those examples as ‘vRAN’ or ‘C-RAN’ as appropriate.

In the coming pages, we will investigate why open RAN matters, what telcos are doing about it – and what they should do next.

Table of contents

  • Executive summary
  • What is the open RAN and why does it matter?
    • Legacy RAN: single-vendor and inflexible
    • The open RAN: disaggregated and flexible
    • Terminology, initiatives & standards: clearing up confusion
  • What are the opportunities for open RAN?
    • Deployment in macro networks
    • Deployment in greenfield networks
    • Deployment in geographically-dispersed/under-served areas
    • Deployment to support consolidation of radio generations
    • Deployment to support capacity and coverage build-out
    • Deployment to support private and neutral host networks
  • How have operators deployed open RAN?
    • What are the operators doing?
    • How successful have deployments been?
  • How are vendors approaching open RAN?
    • Challenger RAN vendors: pushing for a revolution
    • Incumbent RAN vendors: resisting the open RAN
    • Are incumbent vendors taking the right approach?
  • How should operators do open RAN?
    • Step 1: Define the roadmap
    • Step 2: Implement
    • Step 3: Measure success
  • Conclusions
    • What next?

Enter your details below to request an extract of the report