Edge to Cloud Continuum

Context of Edge Computing

Edge means different to different teams/people. Hence it is important to know the context and domain when we talk about Edge Computing.

The edge spans anywhere between the end user device(s) and the cloud/internet. The advantages of edge computing include consumers and enterprises being able to run low latency applications with better performance and reliability, as well as process data close to the data source to reduce backhaul traffic volumes and costs associated with data movement to Cloud and vice versa.

Types & Characteristics of Edge Computing

Broadly these are the types of Edge/Edge Computing frequently referred as relevant to the various industries or businesses in context:

Edge Compute needs to satisfy the different type of application needs pertaining to eMBB, uRLLC and MMTC such as

• Compute Process Intensive (AI/ML or Analytics Workloads)
• Throughput & Integration Intensive (High Data Throughput/Volume/Size, Distributed Data Processing among Edges and/or with Cloud Integration)
• Security Intensive (Sensitive/Highly Confidential Data Processing, Data Encryptions & Decryptions, Data Sovereignty, etc.)
• Latency Intensive (Health care applications, Emergency Services, Autonomous Vehicles, AR/VR, Video Surveillance, Real Time Image Processing, etc)
• Content Intensive (Video Streaming, Data/Event Streaming, Voice traffic, Web traffic, etc.)
• Uplink Intensive (Predictive/Proactive Analytics, Central Monitoring, etc)
• Downlink Intensive (Mobile Apps, Web Traffic, Video Streaming, etc)

Device/User Edge Computing or End Point Computing (On-device)

• This means data processing & application execution at the end user device itself
• User/Device edge represents a highly diverse combination of resources. The closer that edge compute resources get to the end users’ physical world, the more constrained and specialized they become
• In this case, the end user devices will be specially designed and built using purpose built chipset, compute and networking resources as suitable.
• The Applications using this edge computing will have varying characteristics such as Compute Intensive & Latency Intensive in most of the cases and they require Heavy Local Processing with limited or no Downlinking and Uplinking at the edge in most of the cases.

Consumer Edge Computing (Public/Citizen Services Edge)

• This means the data processing & applications execution at the broad set of general purpose consumer devices and/or systems
• Govt Citizen Services, Smart Roads, Smart Cities, Smart Homes, etc where the general purpose devices and systems are used which will have capabilities to carry out generally available application needs.
• Consumer Products and Services offered by various businesses regionally or locally for retail customers using digital platforms
• The Applications using this edge computing will have varying characteristics such as Compute Intensive, Latency Intensive, Security Intensive, Throughput & Integration Intensive in most of the cases and they require Heavy Local Processing with heavy or limited Downlinking and Uplinking at the edge as per the usecase.

Network Edge Computing or MEC (aka Telco Edge or Service Provider Edge)

• Multi-access Edge Computing or Telco Edge or Network Edge or Service Provider Edge is the edge computing offered by Communication Service Providers (CSPs) at the edge of their network to consumer and enterprise customers to connect the devices/applications to the Cloud and Internet easily and reliably.
• Network Edge provides services over the global & regional fixed networking infrastructure. Most of this is leveraged from fixed and private networks developed by the CSPs, many of whom are deploying both Edge services and 5G services.
• Telco edge compute can provide capacity to handle sudden spikes in workloads from unplanned increases in end-user activity or address enterprises’ need to scale quickly when developing, testing and deploying new applications.
• For mobile applications, edge compute not only needs to scale up and down, but also move across different telco edge locations.
• There are multiple potential locations for telco edge computing on and off the public network. These include customer premises, cell towers, street cabinets, and network aggregation points in the access and core network.
• The Applications using this edge computing will have varying characteristics such as Compute Intensive and Throughput & Integration Intensive in most of the cases and they will require Heavy Local Processing with Heavy Downlinking and Uplinking. In this case, there will be digital and service platforms provided at the network edge by the service provider to provide connectivity orchestration and distributed computing platform rather than end user applications.

Mobile Edge Computing

• This means the edge will be moving across the locations in which case the devices and applications will be requiring varying needs in terms of connectivity, latency and throughput aspects
• The devices will be temporarily out of internet connectivity as well depending on the service provider mobile services coverage
• These Edge Applications are mandatorily depend on the roaming services, hybrid connectivity services, multi-mobile coverage plans, etc
• Applications related to Smart Transport Services, Smart Logistics, Smart Supply Chain services, etc will use this type of edge computing.
• The Applications using this edge computing will have varying characteristics such as Compute Intensive in most of the cases and require Heavy Local Processing with limited or no Downlinking and Uplinking at the edge in most of the cases.

Enterprise Edge Computing (Private Edge)

• This means Private Computing as required for Enterprises & Industries, typically Industry 4.0 usecases
• Applications related to Smart Factories, Manufacturing Facilities, Assembly Lines, Factory Automations, Predictive Analytics for Machines/Product Lines/Quality Conformance, etc. require private edge computing within the premises or near the group of devices/machines/equipment.
• Depending on the usecase, these can be Off-premise Edge (Oil Rigs, Ports, etc) and as well as On-Premise Edge (Central Office, Branch Office, Data centers, etc)
• For this type of edge computing, Enterprises are opting for Private Networks using Fixed Fiber, 4G LTE, 5G, Wifi, NB-IoT, Hybrid/Multi Networks, etc. for connectivity solutions and services as per the usecase and application demands
• The Applications using this edge computing will have varying characteristics such as Compute Intensive, Latency Intensive, Security Intensive, Data Traffic & Integration Intensive in most of the cases and will require Heavy Local Processing with limited Downlinking and Uplinking at the edge in most of the cases.
• Industry 4.0 Usecases require Digital Transformation in Industries to turn OT data into actionable insights to improve financial performance and business value (Convergence of IT + OT + CT to generate accelerated business value for the Industries) such as
  - Intelligent Asset Optimization (get most out of highly expensive assets in the factory)
  - Enterprise Operational Intelligence Analytics
  - Scalable Production Management (PLM)
  - Digital Workforce Productivity (replacing of ageing manual workforce in industries using AR, VR, etc)
  - Robotics and Automated Operations for Assembly/Production Lines to reduce the human resources cost and increase the efficiency and quality of products

Cloud Edge Computing

• This means the computing services provided by cloud providers near to the end users for reducing the latency of the data traffic between users and cloud
• In this case, the data processing happens at the Cloud centrally but the data distribution happens near the edge where the data is consumed by the end users/end user applications
• CDNs and Regional/Local Edge Locations provided by Cloud providers belong to this type
• The Applications using this edge computing will be Content Intensive in most of the cases and require Heavy Downlinking with limited local processing at the edge in most of the cases.

Factors Influencing Edge Computing

• Connectivity: devices will not be always connected, unreliable connection, its not “always on” but it will be “anytime on”
• Security: it cant be aftermath thought, it has to be built-in aspect but not bolt-on
• Size & Form factor: it is major factor for many industrial and consumer iot devices where size and form factor matters the most above all.
• Resources Availability: economy matters will be critical for many consumer IoT usecases and it will have direct impact on hardware resources availability for compute and storage
• Upgrades & Maintenance: how frequently the compute software can be upgraded and maintained and can it be done remotely or onshore always. Will the hardware and device itself will be adaptable and affordable for continuous upgrades of software
• Energy: many edge devices will run on battery or limited energy resources and hence the compute workloads have to be highly efficient and environment conversation becomes the key factor
• Volume: Few usecases involve hundreds and thousands of devices or even millions of devices if it is related to smart city or smart governance. In such case managing, maintaining and upgrading of compute will be highly challenging
• Geographically Distributed: for certain use ases devices will be spanned across the geographies and even across continents. In that case compute and workloads have to be able to function independant enough
• Regulation and Policies: IoT devices will involve many regulatory and policy rich conditions due to obvious nature of their existance in consumer perimeter and as well direct public and private policy impacts
• Data Security, Privacy and Protection: Data in transit and at rest has to be protected as per data classification the edge devices handle and also privacy will be the biggest factor in protecting while the device is functional, when it is decommissioned and when it is not functional as well.

Edge Computing Benefits

The cloud offers benefits related to infrastructure cost, scalability, high utilization, resilience from server failure, and collaboration. Whereas Edge computing offers faster response times, lower bandwidth costs and resilience from network failure.

The key benefit of edge computing for enterprises and consumer applications is sending less data back to the cloud for storage and processing. This is particularly relevant for data-sensitive applications and those generating huge data volumes.

By processing data at a network’s edge, edge computing reduces the need for large amounts of data to travel among servers, the cloud, the internet and devices or edge locations to get processed.

• Local Computing with Lower latency: Data processing at the edge results in eliminated or reduced data travel. This can accelerate insights for use cases with complex AI models that require low latency, such as fully autonomous vehicles and augmented reality.
• Less Expensive Data Movements: Using the local area network for data processing grants organizations higher bandwidth and storage at lower costs compared to cloud computing. Additionally, because processing happens at the edge, less data needs to be sent to the cloud or data center for further processing. This results in a decrease in the amount of data that needs to travel, and in the cost as well.
• Remote Computing: Internet access is a must for traditional cloud computing. But edge computing can process data locally, without the need for internet access. This extends the range of computing to previously inaccessible or remote locations.
• Data Security & Sovereignty: When data is processed at the location it is collected, edge computing allows organizations to keep all of their sensitive data and compute inside the local area network and company firewall. This results in reduced exposure to cybersecurity attacks in the cloud, and better compliance with strict and ever-changing data laws.

Edge to Cloud Continuum

Edge computing is similar to data center/Cloud computing as:

• It includes compute, storage and networking resources.
• Its resources may be shared by many users and many applications.
• It benefits from virtualization and abstraction of the resource pool.
• It benefits from the ability to leverage commodity hardware.
• It uses APIs to support interoperability.

Edge computing differs from computing in large Centralized Data centers/Cloud Computing as:

• Edge sites are as close as possible to end users. They improve the experience over high latency and unreliable connections.
• May require specialized hardware and compute capabilities such as GPU/FPGA platforms for AR/VR functionality, AI/ML Acceleration Capabilities, Connectivity Capabilities, etc
• Edge can scale to large numbers of sites and nodes highly distributed in distinct locations or spread over large areas/fields
• Identity of the Edge Devices, edge site’s location and the identity of the access links it terminates are significant.
• Edge sites will join and leave the pool of infrastructure over time. The entire pool of edge sites can be considered to be dynamic. Because of their physical separation, edge sites will, in some cases, be connected to each other and the core with WAN connections.
• Edge sites are remote and potentially unmanned, and therefore must be administered remotely. Tools need to support intermittent network access to the site.
• Edge supports large differences in site size and scale, from data center scale down to a single device.
• Edge sites may be resource constrained; adding capacity to an existing site is restricted due to space or power requirements.
• Isolation of edge computing from Data center/Clouds may be required to ensure that compromises in the “external cloud” domain cannot impact services at the edge.

Edge computing consists of both the edge site (e.g. the compute, network and storage infrastructure and the devices), but also the applications (workloads) that run on it. Any applications in an edge computing environment could potentially leverage any or all of the capabilities provided by a cloud (compute, block storage, object storage, virtual networking, etc) alongside the Computing capabilities provided at the edge itself.

Edge Computing consists of different layers of management, services and capabilities as shown below.

Key Technology Capabilities required for Edge Computing

• Managed Infrastructure Services (Compute, Storage and Networks)
• Managed Data Services
• Managed Observability Services
• Build & Run Cloud native Applications for the Constrained Environments
• Managed Regulatory and Security Services
• QoS and QoE Assurance Services (Low Latency & Performance)

The services that are seeing commercial success are primarily those that support customers to deploy and manage workloads and the underlying infrastructure across a distributed cloud, i.e. cloud and edge, as opposed to only focusing on the edge. Many edge applications are underpinned by container-based infrastructure, sometimes requiring the enterprise to migrate from traditional servers to Kubernetes. These containers need to be deployed easily and scaled across different sites. Vendors that can offer to do this seamlessly across different environments are getting traction from enterprises.

5G MEC Enablement for Edge Computing

Telco Edge aka MEC deployments will enable the developers, retailers, enterprises, citizen services, etc to leverage the different capabilities as shown below

The next-gen network required for edge computing will deliver improvements that enterprise computing requires for all aspects of their network: reliable, robust performance, lower latency, better security and privacy, provisioning agility and lowered costs.

There is significant overlap in the use cases with 4G and 5G such as AR and VR, autonomous cars, industry 4.0, IoT etc. Although edge computing supports these low latency applications, 5G enhances it by improving throughput and reducing latency.

In order to satisfy the Edge Computing needs and the various emerging application needs, it requires improvement in MEC latency, ranging from 5G standalone upgrades, establishing a test regime for more accurate measurement across target markets, and choosing between a private MEC and a public MEC deployment.

Edge Computing Ecosystem

• Facility: The physical site that includes the land/location for the edge data centre (e.g. area around mobile operator’s cell tower), the data centre itself, power and cooling to support it and additional services to maintain and operate the site.
• Hardware: This includes the hardware inside the data centre (racks, servers, processors and the maintenance and operators for these) as well as end-devices.
• Network: Connectivity infrastructure to and from the edge site, as well as traffic routing controls and types of networks to optimise the delivery of content (e.g. CDN).
• Edge Cloud Infrastructure: Virtual infrastructure supporting the edge workloads and applications, from the operating system, the virtualisation layer (which may be container-based), and the platforms for developers to access and manage the storage and compute infrastructure.
• Application/Software: Applications that run on edge computing infrastructure, including network functions, and the application-specific tools that support these, for example analytics capabilities or APIs and platform-as-a-service products.
• Integration & Services: Services that provide support to the customer employing and integrating edge computing at any stage of the value chain — including design and engineering services to create platforms for edge computing applications, or more traditional integration into existing (enterprise) systems.
• Open Source & Forums: Communities that seek to accelerate edge computing — either by creating forums for discussion across stakeholders and industry partners or open platforms to enable developers to build technology.

The edge computing ecosystem is still nascent and one of the challenges is that there are players from different ecosystems coming together: the service providers from the networks, cloud and data centre providers, traditional enterprise IT and industrial applications and systems.

The Hyperscalers have a strong established presence in the cloud already and have already made some movements towards the edge via new products and services such as Azure Stack, AWS Outposts and their IoT offerings. We argue that the edge cloud cannot be standalone and will be an extension of the cloud on a distributed cloud continuum, therefore established cloud providers will undoubtedly be important in this ecosystem.

However, the edge is not homogenous and different use cases may have specific requirements which will be served by others — whether they are industry-specific, or more localized offerings. Plus, there are new capabilities that will be required, for example orchestrating workloads across different clouds, optimizing networks for this distributed architecture and developing applications that benefit from edge computing.

Analyst Insights

• Analysis Mason, a management consulting firm, forecasts cumulative six-year enterprise spending on business applications that require 5G totaling $20 billion USD over the 2022–2027 timeframe, growing at a 75 percent compound annual growth rate (CAGR)
• IoT Analytics estimates that there are now 21.7 billion active connected devices worldwide. 54% (11.7 billion) of these are IoT device connections. By 2025, it is expected that there will be more than 30 billion IoT connections, or almost 4 IoT devices per person on the planet. The trend of connected “things” (iot) is projected to grow 13% annually in next 5 years and by then represent 80% of connected devices by 2025.
• According to market research provided by IDC in 2022, “Worldwide spending on edge computing is expected to be $176 billion in 2022, an increase of 14.8% over 2021. Enterprise and service provider spending on hardware, software, and services for edge solutions is forecast to sustain this pace of growth through 2025 when spending will reach nearly $274 billion.”
• IT leaders tend to agree that 5G is not a standalone solution — in fact, four out of five agree that 5G needs edge compute more than edge compute needs 5G (Figure 41). 5G offers compelling benefits, but it needs an underlying fiber network to deliver the security, availability and TCO that IT leaders need.
• Most of the short-term edge opportunities were in gaming, video analytics, and augmented reality/virtual reality. Though they require low latency, Spirent and STL Partners say the current environment does not support their needs until consistent low latency is offered by 5G edge solutions. Ultimately, latency must be managed holistically, and end-to-end, to achieve reliable and desired customer experiences, and to meet SLAs.
• Nearly 75% of executives consider edge computing a strategic investment, in part because the cost of bandwidth and a centralized infrastructure is seen as prohibitive. Fifty-four percent were comfortable turning to a technology service provider to deliver edge solutions.
• As per the survey [3], more than half of the edge customers surveyed (56 percent) said they would pay for a service-level agreement (SLA) with guaranteed latency that never exceeds a predefined window. However, about two-thirds (66 percent) needed latency of 50ms or less, and the window needed most (37 percent) was 20 to 50ms of latency. That figure contrasts with a global survey conducted by analyst firm IDC which found that 75 percent of business leaders participating in the survey responded that they needed latency of 5ms or less for edge initiatives. To get consistent results in the 5ms range (end-to-end, from device to air interface and network to MEC stack), firms would need to deploy private 5G MEC on premises.

References

1. What is Edge Computing: https://www.openstack.org/use-cases/edge-computing/cloud-edge-computing-beyond-the-data-center
2. ETSI MEC: https://www.etsi.org/technologies/multi-access-edge-computing
3. Edge Computing Latency Demand Survey: https://www.spirent.com/assets/report-cutting-through-the-edge-computing-hype
4. Edge Computing Trends: https://www.lumen.com/en-in/edge-computing/edge-trends.html
5. Global 5G Evolution Summit — Cloud to Edge Continuum: https://youtu.be/ELePdM9g4FM