Can Edge computing enable international carriers to build a global cloud platform?

Can Edge computing enable international carriers to build a global cloud platform?

The Delta Perspective

Authors:
Joao Sousa - Partner
Sam Evans - Partner


Introduction

Edge computing is emerging as a new approach to network architecture that can offer enhanced performance versus traditional Cloud applications without needing to be purely on-premise and quickly becoming a core element of the Amazon Web Services (AWS), Azure and Google portfolios.

In this paper, we assess the market drivers that have supported the emergence of Edge computing and the strategies undertaken by the global cloud providers.  We then address how Edge computing, given its infrastructure requirements, could provide telecoms operators with a new opportunity, if they can successfully create scale through collaboration, to develop a new form of global platform and regain ground in the cloud space.

Understanding the emergence of Edge

The proliferation and advancement of connected devices is driving the greater need for real time processing and storage closer to the customer, avoiding transit latency back to existing cloud platforms.  The advancement of Edge is driven by three main forces addressing limitations of existing architectures, namely:

  1. Appearance of advanced use-cases that require one or multiple enablers, including:
    • Ultra-low latency;
    • Ultra-high reliability;
    • Extreme density of connected devices;
    • Significant processing power at a reasonable cost;
    • Storage and processing of data close to the origin.
  2. Needs of global cloud providers, including:
    • Extending cloud capabilities to device, controller and aggregator level, minimizing dependence on the key limiting factor – connectivity;
    • Improving QoS of existing cloud solutions to address new use-cases and secure market leadership.
  3. Needs of telecom operators, including:
    • Additional mechanisms to monetise assets through addressing emerging Edge computing use-cases;
    • Optimising network economics, especially for 5G deployments;
    • Significantly increasing density of radio access subsystem by moving network functions from existing mobile sites;
    • Reducing the need to reach the core of a network by reducing the number of Single Point of Failure (SPOF) and increasing density of connected devices;
    • Ensuring real time access to the RAN, enabling context and location awareness;
    • Enabling provisioning of advanced use-cases through introducing radio QoS, security, multi-tenancy and low latency.

Exhibit 1: Edge computing landscape: addressing the needs of both global cloud providers and Telcos


These three drivers are rapidly changing the established architecture by increasing the role of devices and aggregation as well as simultaneously introducing a new layer of telco network Edge, an important part of Edge computing.

Edge computing processes data in proximity to where it is produced instead of sending it back to be processed in the Cloud.  By reducing the amount of data flowing to the Cloud, one can significantly reduce bandwidth costs, cloud computing service costs, and dependency on connectivity requirements.  A further key benefit is the improved data security and compliance with data residency laws – i.e. concerns regarding data storage in a country or region different to where the data owner resides.  This is particularly important for specific enterprise verticals as well as government. Currently many countries, especially in emerging markets, have limited local data center infrastructure and so the Cloud is hosted outside the domestic market.

Exhibit 2: Edge computing: enabling advanced 5G use-cases

Despite the many benefits of Edge computing, it will not fully replace the traditional Cloud. This is primarily due to the preferential economics and greater storage and processing capabilities of the Cloud versus Edge. As such, we expect that the future solution will be a mix of pure cloud, pure Edge and hybrid solutions which offer different combinations of cloud vs. Edge benefits optimizing the overall solution cost.

Examples of Edge computing use-cases

Use-cases will evolve from simple “enhanced cloud”, i.e. default cloud services with improved QoS, to fully integrated hybrid and pure Edge computing solutions.

With respect to Edge, global cloud providers and telecoms network operators are developing use-cases with a focus on enabling services that cannot be served under existing network architecture. The service focus is on unlocking three enablers:

  1. Low latency – by bringing compute and storage infrastructure closer to the end-device and avoiding long-distance data transfers to major data centers the customer can experience low single digit latency (versus 20-150ms offered by global cloud providers today) via fixed (fiber) and mobile (5G) access networks;
  2. Security and Compliance – by increasing the number of Edge processing points, there will be fewer occurrences where sensitive data is sent abroad or to a different region, allowing for higher compliance with data localization laws and data security requirements;
  3. Network utilization and reliability – distributed nature of Edge will optimize backhaul and backbone use as well as reducing the number of points of failure, allowing for more robust service provisioning in all environments. This also reduces the global cloud providers dependences on connectivity offered by telecom providers.


To demonstrate how telecoms network operators and global cloud providers are starting to address Edge opportunities, several examples can be highlighted.

Telstra

In 2018 Telstra decided to close c.2,500, around 50%, of its local exchanges and redeploy other facilities as local data centers. Telstra is also assessing multiple use-cases that can benefit from Edge computing such as in the mining and cloud gaming (Exhibit 3) sectors. They have investigated potential opportunities with hardware manufacturers and game developers to address consumer market needs. Cloud gaming consists in virtualize graphical processing in a cloud and replace expensive equipment (e.g. personal computers, consoles) required on the gamer side. By moving hardware to the telecom network Edge, gamers will have the potential to access high-performing computing power with low latency to enhance their gaming experience at a lower cost. Single digit latency of Edge computing is required to ensure the right customer experience.

Exhibit 3: How Edge use-cases leverage latency, security and network utilization benefits

Telefonica

Telefonica has developed a comprehensive Edge strategy, relying upon its OnLife program consisting in the upgrade of existing central offices to Edge data centers. It evaluated multiple Edge use cases, with a focus on use cases with potential short to medium-term impact, such as network storage or gaming. It is also working on evaluating potential partnership with AWS, where Telefonica’s facilities could be leveraged by AWS Greengrass Edge-computing service. 

Telefonica has launched an Enterprise Private Network solution (Exhibit 3) with Ericsson’s technology that brings Telefonica’s communications networks to end customers through an approximation process, called Ultra-Edge Computing, that consists in the UNICA-based network (Telefonica’s cloud global platform) Core reaching the facilities of companies themselves. This approach increases bandwidth significantly, while simultaneously minimizing response times, thus achieving a significant improvement in the communication quality.

Vodafone

Vodafone’s Edge computing trail spans multiple use-cases, revolving around Multi-access Edge Computing network architecture. For the consumer segment, cases include transportation, online entertainment, security; for enterprise, MEC-based communication and industrial solutions. Most cases were trialed as partnerships (together with vendors including Ericsson and Nokia).  Focusing on mobile Edge computing for video streaming (Exhibit 3), Vodafone was able to prove substantial benefits to use MEC to enhance customer viewing online. This use-case was relevant as video is expected to contribute to 75% of data consumed on mobile by 2022 and 4 over 10 viewers drop out video viewing in case of buffering issues.

Vodafone and its partners developed a laboratory environment recreating typical network congestion and latency experienced by mobile users. A MEC platform was in the backhaul aggregation layer of the radio access network, and a local ‘cloudlet’ was used to operate virtual video server. This was compared to video stored on AWS servers. Results concluded that video on the Edge started quicker, with no stalls and lower waiting times due to re-buffering.

Global Cloud providers: Amazon Web Services, Microsoft Azure and Google Cloud

The leading global cloud providers have a clear understanding of the performance (latency, time for processing) and independence (lesser reliance on telco connectivity) benefits made possible by Edge computing for their cloud services. To reap these benefits, they are expanding their current cloud service portfolio to make them Edge ready, covering all network levels – from device to global cloud. IoT and Edge storage are the first service areas to become Edge ready and are commercialized. Of the three, AWS and Microsoft Azure have the most comprehensive IOT offering including Edge, while Google, with less features on IOT, is focusing on building leadership on the entire AI vertical.

AWS was the first global cloud provider to migrate to the Edge.  First, AWS has developed a strong IoT Edge proposition consisting of AWS IOT Greengrass, which is a software that extends cloud capabilities to Edge devices. It also launched AWS Greengrass ML Inference which allows customers to deploy machine learning (ML) models on AWS Greengrass devices.  It also launched AWS Snowball Edge (Exhibit 3), which is a data migration and Edge computing device, suited for local storage and large-scale data transfer. Snowball Edge Compute Optimized provides compute power for use cases such as advanced ML and full motion video analysis in disconnected environments. Customers can use these two options for data collection, ML and processing, and storage in environments with intermittent connectivity (such as manufacturing, industrial, and transportation) or in extremely remote locations (such as military or maritime operations) before shipping it back to AWS for data storage on the cloud.

Exhibit 4: Comparison of global cloud provider Edge solutions

Microsoft Azure followed AWS into Edge with the launch of IoT Edge (Exhibit 3) that is made up of three components:

  1. IoT Edge modules are containers that run Azure services, third-party services, or clients’ code.  Modules are deployed to IoT Edge devices and execute locally on those devices;
  2. The IoT Edge runtime runs on each IoT Edge device and manages the modules deployed to each device;
  3. A cloud-based interface enables clients to remotely monitor and manage IoT Edge devices.


Additionally, Azure announce Cognitive Service containers (Exhibit 3) that allow developers to easily add cognitive features—such as object detection, vision recognition, and language understanding—into their applications without having direct AI or data science skills or knowledge. By deploying Cognitive Services in containers, customers can analyze information close to the physical world where the data resides, to deliver real-time insights and immersive experiences that are highly responsive and contextually aware. Cognitive Services containers that customers deploy on their own. With the launch Cloud IoT Edge and Edge TPU by Google in 2018, all three major cloud players started providing ML service on the Edge. 

Google Edge positioning is especially focused on the development of the whole AI ecosystem (Exhibit 3:) which revolves around the Edge TPU and the Cloud IoT Edge.  Google IoT offering is seen as less comprehensive than that of Azure and AWS, but their ML knowledge is perceived as very high, with a clear strategy of vertical integration.  Specifically:

  • the Edge TPU is an ASIC chip designed especially for running ML applications on the Edge, allowing customers to run compute intensive AI-related workloads. The Edge TPUs are the chip where algorithms carry out the task they have been trained to do. It makes Edge TPUs different from first Google TPUs, which were optimized to conduct the training part of the process. 
  • the Cloud IoT Edge is a software platform that powers Edge computing devices such as robots, turbines, oil rigs. 

A good example of implementation is provided by the company LG, which used both Edge TPU and Cloud IoT Edge to run algorithms in smart manufacturing LCD panel inspection processes, which increased to 99.9% accuracy of processes.

AWS, Azure and Google employ similar principles in their Cloud-to-Edge expansion. They strive to develop an end-to-end technology stack across cloud and Edge to simplify app development and allow for efficient cloud and Edge management. In Exhibit 6 below, it is demonstrated how AWS IoT product portfolio fits along the network stack.

As more cloud services become Edge-ready, the diversity at the Edge will grow. To accommodate this diversity, AWS and Azure will continue to invest and grow their ecosystem of software vendors, system integrators and solution providers. The larger this ecosystem, the more likely they will have the “right” Edge solution, and the more likely they will win over the market.

From the analysis of use-cases and strategies by telecoms network operators and global cloud providers we identify three key points for consideration:

  1. Defining the ‘Edge’ – Global cloud providers and telecom operators use different definitions of “Edge”. For Cloud providers Edge extends up to end-user device levels (smartphone, IoT sensor, laptop, etc.), while for telecom operators, the line is drawn at existing access locations (e.g. central offices, BSC/RNC, etc.) or corporate client premises. Currently, the main challenge faced by global cloud providers is the availability of connectivity infrastructure. Moving Edge/hybrid solutions towards the end-user device level will help them to become less dependent on Telcos. 
  2. Cloud providers adding Edge to the existing portfolio – Global cloud providers are actively adding Edge components to their existing product portfolio. They are leveraging their dominating position in the application market to offer new Edge services or improving and complimenting existing services with Edge/hybrid benefits. This allows them to better position for the future retail market of Edge.
  3. Telcos’ 5G focus – Telecom operators are mostly focused on 5G implications of Edge and seem to be in an earlier stage of retail product development. Telcos are in the best position to become leaders in the development of enablers for 5G Edge use-cases.

Exhibit 5: Overview of AWS Snowball Edge network architecture

Exhibit 6: Overview of AWS IoT stack

Creating value at the Edge

Edge computing solutions utilize components across the entire network technology stack, from global cloud to end-user devices and sensors.  There is a significant difference in terms of accessibility of different Edge components for global cloud providers and telecoms operators due to the nature of their business models and legacy investments. While global cloud providers dominate at the beginning (global platforms) and end (devices) of the service provisioning path, telecoms operators have a strong position in network infrastructure at national and access levels – particularly important for Edge computing is access network connectivity and Edge (e.g. central offices and aggregation points). Global cloud providers’ ownership of network infrastructure is limited and considered as a key bottleneck for expansion from cloud to the Edge, as mobile and fixed access networks require significant investments of money and time to be built.

This situation creates challenges and opportunities for both global cloud providers and telecom operators. On one hand, global cloud providers are more advanced in terms of product readiness and access to end-users. On the other hand, they lack two critical components for a full-stack Edge Computing service: access connectivity and access network Edge (aggregation points).

The value chain of Edge computing has many similarities with traditional clouds. The main difference is the number of additional network levels involved: devices; on premises aggregation/compute; and telco network Edge. The value of Edge Computing generated for service, infrastructure, hardware and software providers are shown at three technology layers:

  1. Applications. Comprehensive consumer, business and telecom solutions addressing end-user needs. This layer includes pure Edge and hybrid (combining four components: device Edge; customer premises Edge; telecom network eEge; and global/national platforms) services that are based on other layers of the value chain;
  2. Enablers. Edge and hybrid (Edge + global) IaaS and PaaS solutions, Edge NFV (mostly for 5G), SDN / SD-WAN, Devices / On premises equipment. This layer includes hardware and software that will significantly decrease time-to-market and required investments for deploying Edge Applications;
  3. Basic infrastructure. Includes basic access network Edge hosting services (such as access to premises, power, a/c, security, etc.) and multiple levels of connectivity – from international backbone to last-mile mobile and fixed access.

Exhibit 7: Infrastructure landscape of Edge

Wholesale Edge: an opportunity or carrier dilemma?

For telecoms network operators there are opportunities to both retail and wholesale Edge computing.  From a retail perspective, based on the likely use-cases there will be demand from specific enterprise verticals, while from a wholesale perspective demand is likely to come primarily from the global cloud providers, who wish to leverage the local telecoms infrastructure to offer Edge solutions to their customer bases. The key question for telecom operators is whether to focus purely on a retail offering, or also to open their Edge capabilities to the global cloud providers, which in turn can lead to greater competition for their enterprise segment.

Given the scale of global cloud platform developer bases (e.g. AWS, Azure), it is likely that developers will wish to develop on their existing stacks as they move into the Edge computing environment rather than develop on new technology stacks, especially at a localized level, for Edge services.  As such, at a retail level, telecom operators will likely to be able to identify specific Edge opportunities, a significant share of the retail market will be served by global cloud providers. Telecom operators can play a major role in establishing Edge Computing environment and creating value by serving the global cloud providers also. They have the ideal network footprint in place to provide basic infrastructure services located at telecom network Edge. 

Exhibit 8: Existing infrastructure challenges and opportunities

Exhibit 9: Layers of Edge value chain

Global cloud providers are actively expanding their data centers and reducing the gap with telecom operator’s network Edge infrastructure – this is largely true in developed markets. A good proxy for understanding the number of telco network Edge data centers required for global coverage will be the number of existing telecom central offices – thousands of locations across the globe. In comparison, the Amazon, the market’s leading cloud provider has 116 Edge DCs globally.

There is an opportunity for telecoms operators to enable global cloud providers expansion from the cloud to the Edge by offering telecom network Edge services such as Edge and Hybrid IaaS/PaaS, Edge NFV and SDN/SD-WAN. Telcos can also self-provision solutions for 5G enablement and unlock value by sharing these solutions with other operators. This approach is more challenging with the ongoing competition from global cloud providers but will allow telecom operators to benefit from cloud provider’s established client relationships and promote a sustainable economic model of Edge computing.

The addition of Edge will make the landscape of available cloud services and application even more complex. Another potential revenue stream for telecom operators lies in existing ICT space – they can offer tailored Edge and hybrid solutions by leveraging own and global cloud providers’ infrastructure and platforms.

Exhibit 10: AWS global Edge data center footprint


Carrier Edge: the global platform opportunity

Can leading telecoms carriers compete with global cloud providers on Edge computing market? We believe that a globally distributed platform of Edge IaaS/PaaS deployed on existing infrastructure of telecom operators can help leading carriers regain ground in the market.

As mentioned above, carriers have widespread network infrastructure for its existing operations, while global cloud providers mostly rely on major data centers with limited densification in key markets. By creating a single unified platform, telecoms network operators can develop a global enabler for applications and services of cloud providers. 

A much denser data-center footprint will be required for Amazon/Microsoft/Google/Facebook to dominate Edge computing in the same way as they are dominating the global cloud environment. Telecom operators have the great advantage of having key physical infrastructure already financed and built, especially significant in semi-developed and emerging markets. 

Exhibit 11: Three waves of potential Edge opportunities

Collaboration between international telecoms carriers is the key to successfully develop and deploy this globally distributed platform. Apart from individual approaches on exploring Edge computing opportunities, we have identified three priority directions for Telcos depending on the extent of collaboration:

  1. Global network of Edge IaaS/PaaS suitable of providing NFV/SDN for 5G – a single standard covering hardware ecosystem, Edge computing platform, cyber security for the Edge. True 5G is not possible without having Edge infrastructure in place. Standalone deployments are putting significant pressure on operators’ financials. International carriers can jointly deploy the required infrastructure in their preferred countries and reap benefits by sharing it. This can be done through assigning geographical markets to different OpCos and then exchanging infrastructure access. Given the nature of IaaS/PaaS and NFV/SDN/SD-WAN, multiple complexities of sharing the infrastructure will be solved automatically. These deployments will naturally lead to the next opportunity;
  2. Enabling global cloud providers in semi-developed and emerging markets by offering Edge IaaS / PaaS – Main battlefields for global cloud providers are developed markets, this is reflected by the fact that many countries still do not have a single data center. Pre-provisioning of Edge IaaS/PaaS by telecoms operators will drive global cloud providers to consider partnerships over building its own infrastructure. In the most underdeveloped markets, availability of Edge data centers will have major implications on global cloud providers’ decision on entering the market. While in countries where Amazon/Microsoft/Google already have a major data center, enabling Edge applications will significantly improve their time-to-market, allowing telecoms operators to benefit from the extensive retail footprint of global cloud providers;
  3. Long-term opportunities on advanced use of Edge - Constructing unified cloud computing infrastructure and providing service capability with key features of computing, storage, network and security are inevitable choices of telecom operators looking to migrate away from the “dumb pipe” position in the tide of ICT convergence. As IT and CT converge, Multi-access Edge Computing will be critical for the digital transformation for operators, as this helps operators to quickly build cooperation with OTT or application developers. Telecom operators can open the storage and computing capabilities of the MEC platform to application developers and content providers, providing them with a brand-new service development environment and user experience. In addition, the information at the wireless side can be encapsulated into various services (for example, RNIS, location-based service, and bandwidth management service) running on the MEC platform and available to enterprises and vertical industry.

There is already at least one good example of the platform development approach. Deutsche Telekom founded MobiledgeX – a company that is creating a marketplace of Edge resources and services that will connect developers with the world’s largest mobile networks to power the next generation of applications and devices. MobiledgeX is building a developer-friendly, multi-tenant, on-demand set of Edge services to enable the deployment of applications, application components and services much nearer to connected users and devices than is possible with today’s centralized cloud services. From a developer perspective, the goal of MobiledgeX and the Edge ecosystem is to make sure any Edge services offered seamlessly integrate into existing ways of developing and do not add any additional operational burden than already existing in best practice cloud-based design workflows. (source: MobiledgeX website)

Exhibit 12: Overview of MobiledgeX architecture

Video

Delta Partners - TMT Advisory and Investment