What Is a Middle-Mile Network and How Does It Impact My WAN / SD-WAN?

Regular readers may be familiar with the concept of a middle-mile network (we’ve explored this topic in previous articles). As a reminder, middle-mile connectivity lies between the last mile and the greater internet backbone. For many ISPs, their middle-mile network is their “core” network.

But even if you know your last mile from your first mile, you might still be a little hazy about exactly why the middle-mile can make the difference between a fast, reliable SD-WAN, and a slo-mo, juddering car crash of a network.

Different SD-WAN Topologies

As a refresher, there are three main types of SD-WAN topologies (with variations). The importance of the middle-mile to your network is dependent on which topology you’re using.

On-premise Site-to-Site Dynamic VPN Topology

Each site on your network is connected via an encrypted virtual private network (VPN) tunnel that runs from the site’s SD-WAN customer premises equipment (CPE). This is well insulated from the middle-mile network – so far, so good.

Cloud Gateway Topology

This “hub-and-spoke” topology is designed to allow the CPE at all the different sites to connect to a cloud gateway. This cloud gateway either redirects the routing and connects the sites to each other or acts as a host to the connection.

Issues in the middle-mile can affect individual connections in this topology, but with enough redundancy and diversity planning, the majority of middle-mile problems can be anticipated and avoided.

Cloud Network Topology

This topology (also the basis for most SASE topologies) relies heavily on the middle-mile network for the core of its operation.

Here’s a (slightly idealized) diagram of how an enterprise-level cloud network SD-WAN operates.

As you can see, everything’s going through the middle mile – every user, every server, every cloud app, or SaaS instance.

However, to understand the kind of middle-mile issues that might impact this network, we’d better use a diagram that shows the physical locations. If we presume a simplified network that’s restricted to the U.S. only, it might look a little like this.

So, for example, if the L.A. branch were looking to send or receive traffic from the cloud server in Washington (via the northerly route), the stages would run as follows.

• Traffic leaves L.A. branch CPE, via public internet

• Traffic arrives at the nearest cloud gateway/POP and joins the service provider’s middle-mile network

• Traffic is routed by service provider, via gateways in Seattle, Chicago, New York, and Washington D.C.

• Traffic leaves the middle mile, arriving at the cloud-hosted server

In optimal circumstances, this network would offer consistent performance and reliability. With only the last-mile segments of the network being a public internet connection, most traffic transactions will proceed with little variation.

However, if there is any disruption in your service provider’s middle-mile network, the consequences for a cloud networked WAN could be substantial – and potentially affect the entire network.

In the diagram below, we’ve hypothesized a link outage between L.A. and Dallas.

As you can see, it’s not just traffic between those two points that would be affected – the outage would put additional strain on pretty much every segment of the network. 

The potential for bottlenecks (as we can see here between Chicago and New York) would be greatly increased. And access to the cloud-hosted servers in Washington would become more difficult, especially for sites further away (like L.A.).

Protecting your Cloud Network SD-WAN

SD-WAN network providers are generally aware of the risks we’ve outlined here. These services tend to be built with plenty of redundancy and fault tolerance baked in.

However, if you’re in the market for a cloud-network, SD-WAN topology, it’s essential that you account for these conditions in your procurement process. Think of asking the following questions.

• Does your preferred provider’s cloud-network topology offer a robust middle mile?

• Can they provide evidence of their redundancy and diversity (preferably by supplying .kmz files)?

• Are they equipped with robust connections to your preferred current and future cloud service providers?

• Where are their gateways located in relation to your sites of operation? (Again, .kmz files would be useful to answer this question.)

Procuring enterprise network connectivity is an increasingly complex undertaking, due to questions like this (and many more). 

With the Lightyear Telecom Operating System, you can undergo this process with an automated platform that will help simplify and clarify your options. In addition, you are supported by experienced industry professionals who would be happy to chat or meet in-person to talk through your network questions.