The Cable Perspective on Edge Compute and 10G Networks
In recent years, organizations have increased their investment in data centers and cloud computing, pulling network functions into a centralized data center using a cloud computing model for service and delivery. However, the network’s powering, storage arrays, capacity, and compute were still delivered from the data center. As operators realized the need for low latency applications, support for lightweight edge devices was needed and this observation led to the notion of distributed compute at the edge using cloud models. Additional consideration given to the requirements of edge node processing paved the way for edge compute.
The benefits of an ultra-reliable network yield countless benefits for the proliferation of devices that have grown beyond a centralized architecture. Edge compute, or serverless edge computing as a distributed open platform, allows the edge of the network to handle various functions like processing, storage, caching, and applications.
Edge compute represents the future for telecommunications companies. Why should we care? Edge compute in conjunction with a 10G and cloud architecture brings technology closer to the customer, enhancing such technologies as Smart Cities, Internet of Things (IoT), mobility, autonomous vehicles, Telehealth/Telemedicine, and Industry 4.0.
"With greater capacity in the network, operators will need improved methods to crunch the data, scale connectivity for devices, and maintain quality of experience (QoE)."
Nodes Where They are Needed
The edge compute (EC) model allows operators to use lightweight nodes like IoT or mobile to process, analyze, and/or store data at the edge. The idea is to bring processing, data storage, and analysis, along with applications, closer to the data. By making edge compute closer, operators will save bandwidth, reduce latency, reduce jitter, process big data, and improve other key performance indicators (KPIs) in the network. The open question is how to ensure that the data remains close to the customer. EC that leverages 10G has the capacity to do that.
EC is the key technology that can support innovative services for a wide ecosystem, with stakeholders including connectivity operators, infrastructure providers, application, and content providers. When you have smart components out in the access network, close to the customer, EC empowers computation so that it doesn’t have to be taken back to the facility. EC enhances the cloud to stay close to the innovative technology being supported. When EC is delivered in conjunction with a 10G running over virtualized digital networks, it allows software to define the network. With greater capacity in the network, operators will need improved methods to crunch the data, scale connectivity for devices, and maintain quality of experience (QoE).
A growing trend in access network technology today is deploying multi-access edge computing, or MEC. This trend requires a re-architecting of the network edge, with the objective of providing greater computing power closer to the customer. Let’s take a closer look at the components of MEC. (See Figure 1.)
Historically, the cable industry has built its access technology predominantly on hybrid fiber coax, or HFC. Increasingly, access network connectivity provided to customers comes in a variety of forms, including HFC, PON, Wi-Fi, and cellular, both LTE and 5G. Customers just want their services to work across all their devices, wherever they are, regardless of access protocol. The telecommunications industry now has a flexible “playbook” to deploy different access protocols to support different customer wireline and wireless requirements. Laying the groundwork for a converged access network.
Where’s Our Edge?
Where exactly is the edge of the access network? Is it at the hub site, the fiber node, the amplifier, the splitter, the access point into the home, or even the customer’s home gateway? This is a frequently asked question in the telecom industry, with discussions ranging from the architectural to the operational. Defining the edge depends on the perspective of the provider.
A business or enterprise is likely to view the edge as their equipment on-premises. A wireline network operator may define their edge as the piece of active network equipment closest to the customer. A wireless network operator may define the edge as the cellular radio. With such a “soft” definition, it is important to specify your meaning of “edge” before you go too deeply into a service delivery discussion. MEC implementations utilize all of these locations, depending on the service.
Cable Computing
Traditionally, the bulk of computing power in a cable network has been performed in a data center, headend, and/or hub site, miles away from the customer. With MEC, computation and memory are positioned much closer to the customer, considerably shortening the distance, and time, it takes to complete a service transaction, while also lessening the bandwidth needed to transport across the access network.
When you put it all together as Multi-Access Edge Computing, you get a trend that leverages one of the greatest strengths that the cable network operator has: direct connections with millions of customers’ homes.
"Edge compute in conjunction with a 10G and cloud architecture brings technology closer to the customer, enhancing such technologies as Smart Cities, Internet of Things (IoT), mobility, autonomous vehicles, Telehealth/Telemedicine, and Industry 4.0."
With MEC in place, the following benefits accrue to the network operators and their customers:
- Low Latency. Many services, such as gaming, virtual reality (VR), and financial transactions, require predictable and low latency. With MEC, the data remains closer to the customer, greatly reducing latency.
- Reduced Traffic Load on the Core Network. Many transactions can be addressed within the MEC node, without the need to traverse the core network.
- Increased Analytics and Security. Harnessing the processing power of the MEC node, service analytics and enhanced security are performed more efficiently close to the data flow origination.
- Convergence. With computing power at the edge, interworking across access protocols is supported.
- Decentralization. Via the MEC node, the network operator can deploy services to specific customers or groups of customers.
- Open Architecture/APIs. MEC has been developed based on standards and open interfaces, encouraging new and innovative companies to develop services.
Each of the above factors contribute to a faster introduction of new services, as well as better performance and availability to the customer.
The Society of Cable Telecommunications Engineers (SCTE®), a subsidiary of CableLabs®, has been hosting the Generic Access Platform (GAP) working group to define common interfaces within fiber node housings that support the convergence of wireline and wireless by accommodating DOCSIS, PON, Wi-Fi, 5G, and business services. (See Figure 2.) The GAP architecture also supports MEC, with modules dedicated to processors and memory. The GAP standards are an important tool in the push to bring 10G to life. It will increase reliability with redundant, interoperable, and replaceable modules; decrease latency with remote computing power in the node; and increase high-speed communication between interoperable multi-vendor modules.
Moving closer to the smart city or customer is the next logical step. With the increasing volume of IoT and mobile devices, networks need a way of crunching all that data closer to the user, rather than transferring it all the way back to the connectivity provider. The volume of devices is growing exponentially, and the industry is working toward wireless convergence, providing greater connectivity. The possibilities are seemingly limitless and we’re already seeing applications in medical and exercise sensors. EC is necessitated by the emergence of the IoT because processing power must be located near connected devices. The question now is how much infrastructure must be deployed in close proximity to ensure low latency? 10G may provide the answer to the question of capacity.
ABOUT THE AUTHOR
Steven H. Harris is Executive Director, Education and Technical Sales, L&D for SCTE®, a subsidiary of CableLabs®. He has more than 25 years of experience in L&D and IT/telecommunications. For more information, please email: [email protected] or visit: www.scte.org. Follow Steve on Twitter @stevenharrisnj and SCTE on LinkedIn and Twitter @scte and on Facebook: TheSCTE.