Lee Doyle is Principal Analyst at Doyle Research
In network terms, the edge is where the action is. The implementation of 5G wireless networks, connectivity of billions of IoT devices/sensors and low latency edge computing applications are changing the requirements of the telecom edge. The convergence of wired and wireless networks is driving new service offerings for both business and residential customers.
Communications Service Providers (CSPs) are partnering with hyperscale cloud providers to deliver new services at the edge – examples include AT&T with Microsoft and Verizon with AWS. CSPs are hoping to capture new revenues from edge services with the help of the hyperscalers’ cloud presence and customer base. For cloud providers, connecting with CSPs enables them to deliver competitive low latency services based on 5G.
Re-evaluating network architecture at the edge
CSPs of all types are re-evaluating their network architecture at the edge in terms of performance, agility and cost. They have already virtualized their core network applications on powerful x86 servers. Operators now are leveraging network silicon advances (e.g. Broadcom Jericho) combined with innovative networking software to enable new edge services.
A software-driven network powered by cloud-native technologies (microservices, containers) can overcome the limitations of current network architectures at the edge. Edge routing is especially ripe for disruption. Cloud-native, disaggregated edge routing software on merchant silicon can deliver the performance, intelligence, agility and cost characteristics to power converged network edge requirements.
What new applications are driving requirements at the edge of the network?
Edge computing is an architecture that supports distributed compute capacity near its data source. Networking at the edge provides the high speed, low latency connectivity necessary to link the sources of the data (e.g. PCs, PoS and IoT devices) to compute and storage – either at the edge or the core. The network edge is adapting to support the billions of devices and sensors now connected to the Internet – including sensors, robots, smart meters, shipping containers, surveillance cameras, and thermostats. There is a wide range of applications driving new edge network requirements including:
- Video analysis
- AI/ML for enterprise IoT applications
- High performance SaaS applications
- Autonomous vehicles
- Public safety, traffic
The buildout of the 5G wireless network is also driving new network edge requirements, especially to connect the large number of cell sites required to deliver high speed wireless data services.
How do partnerships between telecom service providers and hyperscale cloud providers impact network requirements?
CSPs and hyperscale cloud providers are increasingly forming partnerships to accelerate operations at the edge. The goal is to deliver low latency connectivity to support edge computing operations.
CSP and cloud providers continue to upgrade connections at their transit gateways (to 100GB and up). Additional relationships are established to offer enhanced edge computing (e.g. IoT services) to enterprise customers.
These partnerships will require a new network architecture with a high speed, flexible edge network. The edge network will need to be rapidly and cost effectively deployed at 100s or 1,000 of sites. Each site may have different requirements for capacity, space and power. The edge network must scale from small to large and enable the agile development of new application aware services via open APIs.
Cloud native, containerized network software will enable appropriate scalability and flexibility for the new edge network. Low cost white box will power the thousands of edge locations.
Why are CSPs eager to transform their network at the edge?
CSPs are under pressure to transform their networks to deliver high value services. The alternative is continual cost pressure as new networks (e.g. 5G and FTTH) offer enhanced performance at lower prices.
Edge computing, remote workers (office of one) and IoT applications offer significant new revenue opportunities for CSPs. They must modernize their networks to support edge requirements for low latency and high capacity. For example, 5G networks will offer approximately 5 times the performance of 4G and require 5 to 10 times more base stations.
Their networks are burdened with high cost proprietary network appliances with low resource utilization. They have the operational complexity of operating a number of separated networks (e.g. MPLS, Internet, and wireless) with a plethora of network equipment each with its own operating environment.
The goal for many CSPs is to converge and simplify network design and operations. They need lower capital costs to upgrade their network and make it easier to manage to reduce operating costs. For example, the ORAN initiative is being driven by a number of leading CSPs to dramatically drive down the cost of 5G RAN deployments and deliver open platforms for service innovation. The same is needed for broadband services.
What are the benefits of a software-based disaggregated network at the edge?
To enable service agility and scalability, the edge network will need to be architected like hyperscale clouds – i.e. open, standard, cloud-software based.
Advances in network software combined with powerful new silicon is creating new opportunities for edge network architectures. A number of organizations are driving innovative network architecture design including OCP, Open RAN, TIP, ODIM and DriveNets.
For example, with 5G, mobile operators need to deploy a large number of cell site routers. These routers need scalability, high performance and operational automation. Cloud-native routing on white box hardware can significantly drive down acquisition costs as compared to traditional hardware routers.
Networking software provides the advantages of flexibility and the ability to update software remotely. Its advantages include:
- One architecture can span a variety of network needs – e.g. cell site, peering, aggregation, cloud interconnect, etc.
- Network automation which enable easy update of routing functionality = lower operational costs
- Flexible pricing models
- Open APIs to enable the rapid development of new services
The disaggregated model will drive new edge computing opportunities
A software-driven network powered by cloud-native technologies is driving implementation of a new network architecture at the edge, just like the cloud data center architecture. Disaggregated network software enables CSPs to cost effectively upgrade the performance of their networks, more efficiently leverage network resources and helps them converge multiple networks. It enables streamlined connectivity between network and cloud resources – thus driving new edge computing opportunities.