VNFs on public cloud: Opportunity, not threat

VNF deployments on the hyperscale cloud are just beginning

Numerous collaboration agreements between hyperscalers and leading telcos, but few live VNF deployments to date

The past three years have seen many major telcos concluding collaboration agreements with the leading hyperscalers. These have involved one or more of five business models for the telco-hyperscaler relationship that we discussed in a previous report, and which are illustrated below:

Five business models for telco-hyperscaler partnerships

Source: STL Partners

In this report, we focus more narrowly on the deployment, delivery and operation by and to telcos of virtualised and cloud-native network functions (VNFs / CNFs) over the hyperscale public cloud. To date, there have been few instances of telcos delivering live, commercial services on the public network via VNFs hosted on the public cloud. STL Partners’ Telco Cloud Deployment Tracker contains eight examples of this, as illustrated below:

Major telcos deploying VNFs in the public cloud

Source: STL Partners

Enter your details below to request an extract of the report

Telcos are looking to generate returns from their telco cloud investments and maintain control over their ‘core business’

The telcos in the above table are all of comparable stature and ambition to the likes of AT&T and DISH in the realm of telco cloud but have a diametrically opposite stance when it comes to VNF deployment on public cloud. They have decided against large-scale public cloud deployments for a variety of reasons, including:

  • They have invested a considerable amount of money, time and human resources on their private clouddeployments, and they want and need to utilise the asset and generate the RoI.
  • Related to this, they have generated a large amount of intellectual property (IP) as a result of their DIY cloud– and VNF-development work. Clearly, they wish to realise the business benefits they sought to achieve through these efforts, such as cost and resource efficiencies, automation gains, enhanced flexibility and agility, and opportunities for both connectivityand edge compute service innovation. Apart from the opportunity cost of not realising these gains, it is demoralising for some CTO departments to contemplate surrendering the fruit of this effort in favour of a hyperscaler’s comparable cloud infrastructure, orchestration and management tools.
  • In addition, telcos have an opportunity to monetise that IP by marketing it to other telcos. The Rakuten Communications Platform (RCP) marketed by Rakuten Symphony is an example of this: effectively, a telco providing a telco cloud platform on an NFaaS basis to third-party operators or enterprises – in competition to similar offerings that might be developed by hyperscalers. Accordingly, RCP will be hosted over private cloud facilities, not public cloud. But in theory, there is no reason why RCP could not in future be delivered over public cloud. In this case, Rakuten would be acting like any other vendor adapting its solutions to the hyperscale cloud.
  • In theory also, telcos could also offer their private telcoclouds as a platform, or wholesale or on-demand service, for third parties to source and run their own network functions (i.e. these would be hosted on the wholesale provider’s facilities, in contrast to the RCP, which is hosted on the client telco’s facilities). This would be a logical fit for telcos such as BT or Deutsche Telekom, which still operate as their respective countries’ communications backbone provider and primary wholesale provider

BT and Deutsche Telekom have also been among the telcos that have been most visibly hostile to the idea of running NFs powering their own public, mass-market services on the public and hyperscale cloud. And for most operators, this is the main concern making them cautious about deploying VNFs on the public cloud, let alone sourcing them from the cloud on an NFaaS basis: that this would be making the ‘core’ telco business and asset – the network – dependent on the technology roadmaps, operational competence and business priorities of the hyperscalers.

Table of contents

  • Executive Summary
  • Introduction: VNF deployments on the hyperscale cloud are just beginning
    • Numerous collaboration agreements between hyperscalers and leading telcos, but few live VNF deployments to date
    • DISH and AT&T: AWS vs Azure; vendor-supported vs DIY; NaaCP vs net compute
  • Other DIY or vendor-supported best-of-breed players are not hosting VNFs on public cloud
    • Telcos are looking to generate returns from their telco cloud investments and maintain control over their ‘core business’
    • The reluctance to deploy VNFs on the cloud reflects a persistent, legacy concept of the telco
  • But NaaCP will drive more VNF deployments on public cloud, and opportunities for telcos
    • Multiple models for NaaCP present prospects for greater integration of cloud-native networks and public cloud
  • Conclusion: Convergence of network and cloud is inevitable – but not telcos’ defeat
  • Appendix

Related Research

 

Enter your details below to request an extract of the report

Building telco edge: Why multi-cloud will dominate

Defining the edge

Edge computing will remain a focus for telecoms operators for the foreseeable future, both to optimise the network and enable new, third-party applications and services. In fact, 70% of survey respondents believe investment levels of edge computing for supporting third-party applications will increase over that for internal network infrastructure in the next five years.

This report explores how telecoms operators will build their edge computing business, infrastructure and services, and the role multi-cloud will take in this. Before diving into this, it is worth defining this confusing and complicated space. At a high level, edge computing refers to cloud-native computing (and storage) being brought closer to the end-device or source of the data, rather than centralised in a remote, hyperscale data centre.

The telecoms industry has been exploring the role of edge computing for over four years, starting when network functions virtualisation (NFV) began to make real strides. The initial interest was in mobile edge computing (MEC), but this has now evolved to multi-access edge computing to incorporate fixed networks and non-cellular networks too. Outside telecoms, there is edge compute capacity in regional data centres provided by third parties centres, e.g. data centre operators and cloud providers. These are often in untapped geographies, such as Tier 2 cities. In addition, there is edge compute at customer premises, e.g. business campuses or factories.

We outline the scope of edge computing below. There is a full spectrum of possible edges from devices to regional data centres. Some of these edge locations may be owned and/or operated by communications services providers (CSPs). The CSP edge contains the most relevant types of edge for CSPs: network edge and on-premises enterprise edge. They contain infrastructure either owned by a telecoms operator (e.g. a CSP data centre) or operated by one (e.g. network CPE at a customer site).

Enter your details below to request an extract of the report

The spectrum of edge computing locations

There are two main types of applications that can be processed on CSP edge computing:

  1. Telecoms applicationsthat run, protect and monitor the network – i.e. CSP’s own network functions;
  2. Consumer/enterpriseapplications – which CSPs may provide for third-party customers.

STL Partners has been supporting the telecoms industry in exploring the opportunity to provide services and solutions to third parties by leveraging their edge computing infrastructure. These could include enterprises deploying IT applications locally to comply with data sovereignty laws, developers using edge to optimise their applications, IoT solution vendors using edge to reduce latency for mission-critical applications, etc. Our survey highlighted the importance for CSPs in investing in the infrastructure for these applications. On average, CSPs believe that 40% of edge computing investments in the next 1-2 years will be used to support these applications, rather than be used for network functions infrastructure.

Defining edge computing within telecoms

Although the edge computing market is nascent, there are emerging use cases that seek to take advantage of edge computing’s main benefits. These include offering the flexibility that comes with the cloud more local to reduce latency, improving reliability, keeping data secure, and offloading processing from the end-device. However, use cases are at different stages of maturity; some will be deployed in the next two years in early adopter markets, others are more than five years away from commercial, wide scale deployments.

The maturity stages of edge computing use cases

Telecoms operators are keen to leverage edge computing to grow revenues, particularly in their enterprise business. There are different strategies emerging: one is to focus on enterprise connectivity and networking, another on developing a horizontal, cloud-like platform for developers, while a third focuses on building end-to-end solutions for specific verticals.

Types of edge services and business models

The challenge with any new technology is that it takes time to educate the market and engage the innovators who will build the applications that will leverage its potential. Edge computing is complex, because it has a unique ecosystem that spans several industries: cloud, telecoms, industrial, traditional ICT, plus specific vertical sectors. In order to build an edge-based solution, there needs to be adequate infrastructure (facility, hardware, connectivity, edge cloud) plus the applications and services, and these need to be integrated so they work together seamlessly.

The edge value chain

Regardless of the business model and services strategy a telecoms operator chooses to pursue, it will need to first determine how best to build its edge infrastructure to optimise results. This report will dive into three key questions CSPs are still trying to evaluate:

  1. How should telecoms operators build edge computing infrastructure that can support both enterprise applications and network functions?
  2. To what extent should telecoms operators work with partners, particularly the hyperscalers, to build their edge and take services to market?
  3. How can telecoms operators effectively work with the ecosystem?

Table of Contents

  • Preface
  • Executive Summary
    • There are three key factors to consider to build the CSP edge
    • The edge will be multi-(edge) cloud
    • CSPs must build capabilities and partnerships today to support their edge business
  • Defining the edge
  • Laying down the foundations: Options for building the CSP edge
    • Convergence
    • Organisation
    • Hyperscaler partnerships
  • There is no single edge – it is multi-cloud
  • Conclusions and recommendations: What CSPs should do next
  • Index

Enter your details below to request an extract of the report