An edge network refers to globally distributed infrastructure that offers full-stack infrastructure services, which includes networking, compute, and storage, within 20 milliseconds of major metropolitan areas.

Compared to traditional network services, which only offer backbone connectivity between sites, an edge network exposes the infrastructure used at the points of presence to enable customers to build, deploy, and run custom logic and applications.

The term ‘edge’ as a product category and marketing term evolved from its usage in network architecture; namely the topological edge of the network. Network operators – such as communications service providers – handle their networking infrastructure in two layers: edge and core.

In absolute terms, the edge is the external-facing component of a network.

The edge of the network consists of external-facing networking appliances (i.e. edge routers) that connect external services into the network operator’s backbone and vice versa. Core network appliances handle communications between edge routers and consist of high-capacity switches and links that handle very large amounts of traffic.

With different network architectures and as-a-service consumption models, ‘edge’ became an umbrella term that includes cloud edge, far edge, mobile edge, and near edge – terms which we will explain later in this blog.

As such, the term is used to describe services that offload compute from the end-user device to a dedicated server managed by a third party. All edge architectures will inject some amount of network latency to process requests away from the end-user device. Depending on where the edge server is hosted, network latency will vary. Some calculators can estimate the amount of latency depending on distance, but a good rule of thumb is that 1000 miles will incur 20 milliseconds of roundtrip latency.

Four (Common) Types of Edge 

Considering the above, we can define and classify the following types of edge architectures:

• Cloud edge – the cloud also enters the category of processing requests away from the end-user device. These have mature compute and storage services, but their networking component is limited. A relatively small geographical coverage for their data center estate induces a considerable amount of network latency. Cloud providers are also unable to route requests to the closest datacenter available, and point all requests to a single origin. Induced network latency >50ms.
• Far edge – compared to cloud providers, far edge solutions operate a large number of globally distributed points of presence interconnected by a network backbone. These focus on providing edge services with minimal latency across geographies. Using Anycast, far edge solutions can route requests to the closest point of presence for low network latency and provide global load balancing. Induced network latency 10 – 50 ms.
• Mobile edge – these offer compute services using mobile network operators’ wireless and radio infrastructure. Wireless connectivity is both an advantage and disadvantage, where use cases are typically limited to mobile and IoT devices, and are subject to coverage in the area. Induced network latency ~1 ms.
• Near edge – these are solutions deployed by customers on their premises. These are very similar to traditional client-server models, but have a central management component that removes the friction associated with managing multiple servers individually. This is the only model described here that is not consumed as-a-service. Induced network latency <1 ms.

The four types of edge described above are the most common ones. However, there are some emerging topics that are gaining traction, such as:

• Edge AI – refers to the deployment of AI applications in devices throughout the physical world, where the AI computation is done near the user at the edge of the network, close to where the data is located.
• Terrestrial edge – rather than having compute offloaded to a data center or point of presence, it is offloaded to a satellite device. This is also referred to as a satellite-terrestrial edge computing network (STECN).
• Low-Power Wide Area Edge – LPWAN offers extensive wireless coverage that is often decentralized, with multiple individual users and organizations running gateways to provide coverage in their local area.

In multi- and hybrid-cloud environments, edge networks fill in the gaps left by cloud providers. These include features such as network security, load balancing, network performance, and high availability.

To address network performance, edge networks offer self-service mechanisms to deploy private connections between cloud and on-premises data centers. This can offer a range of advantages, such as deployment flexibility, advanced routing to reduce latency, and cost optimization. 

Edge networks can provide network security by encrypting and filtering traffic, protect against distributed denial of service attacks, and act as a reverse proxy to hide origins.

Anycast networks are a distinguishing feature for edge networks that help with all of the above, namely performance improvements in terms of latency, load balancing for distributing requests depending on usage, and high availability that reroutes traffic to nearby points of presence in case of failures.

Geolocation and compliance is another use case where edge networks can help, such as moving workloads away from the cloud and closer to end-users in cases of data sovereignty. 

Comparing Edge Networks

Edge networks vary widely depending on their featuresets. These can include the following:

• Management – as edge networks are consumed in an as-a-service model, providers allow customers to manage their connectivity via a management plane, which can be a self-service portal. These are used to create, configure, and remove compute and network instances, define security policies, manage costs and the like. Edge network providers also expose services via APIs to allow customers to programmatically define their services.
• Anycast – network addressing and routing method in which multiple servers can share the same IP address, whereby requests can be routed to the closest point of presence for the lowest network latency.
• Speed and connectivity – providers can offer a range of port speeds, provide direct access to major cloud providers, and connect into partner networks for last-mile delivery.
• Reliability – to ensure both performance and business continuity, edge networks can support high availability configuration, failover policies that reroute traffic, load balancing for handling requests during busy periods.
• Monitoring and SLAs – Outsourcing network services to a third party also entails a loss of visibility and control. Using an edge network solution for enterprise-grade connectivity requires consistent 24×7 monitoring and the ability to deliver on SLAs, provides incident response and business continuity services.
• Flexibility – offering full-stack infrastructure services, edge networks can provide a wide range of value-add services, which can include authoritative or recursive DNS, colocation facilities to host customer equipment at points of presence, bare metal servers, distributed denial of service protection, and others.

NetActuate’s Managed Edge

NetActuate’s infrastructure services leverage a highly optimized network that can lower latency to major metropolitan areas to under 20 ms roundtrip. 

Building on top of its interconnected footprint of datacenters, NetActuate exposes infrastructure hosted at its points of presence for customers to define edge logic and host applications using compute services. NetActuate uses Anycast which makes redundant servers available from the same IP. If one location is down for any reason, users are sent automatically to the next closest location.

An integrated network and edge compute means that customers do not need to use two or more providers for network and compute services. As such, NetActuate’s edge network solution is suitable for organizations such as:

• Content delivery networks – to extend their infrastructure footprint to new geographies by either deploying their own servers in a colocation environment or running virtual instances on top of bare metal servers or virtual machines.
• Multinational enterprises – providing private connectivity between branches and office locations, and allowing expansion to new locations with a lower network CAPEX investment.
• Communications service providers – offering a consumption-based mode to expand network capacity.
• Media organizations – helping with global media delivery and adjusting costs for off-peak and on-peak consumption.
• Data center and colocation providers – expanding data center footprints in major metropolitan areas with carrier-grade connectivity.
• Secure Access Service Edge (SASE) providers – extend geographical coverage, load balance requests. and ensure network uptime and performance.
• Managed Security Services providers (MSSPs) – improve the speed and reliability of their real-time analysis infrastructure such as web application firewalls (WAFs) and security information and event management (SIEM) tools.

To learn more about NetActuate’s Managed Edge solutions, schedule a call with one of our engineers.