What Powers the Core of Your Enterprise WAN?

Digital transformation. Cloud migration. Branch simplification. This perfect storm is straining legacy networks and sending enterprises scurrying for new WAN answers.

The good news is there are many solutions emerging that promise to help. The bad news: There are so many tools coming out from a variety of players in different corners of the industry, it is hard to figure out how best to approach this critical need.

So, let’s examine the options, some of which involve a Layer 3 network approach versus a Layer 2 approach:

• Public Internet – The first inclination for many organizations is to simply use the public Internet to augment their existing WAN, which is typically MPLS. While that makes it easy to add bandwidth, it will be obvious in short order that the best effort, Layer 3 public Internet is no substitute for the corporate WAN. There are the obvious security questions and concerns with using the public Internet. When it comes to performance, experiences will vary all over the map as congestion, packet loss, latency, and jitter take their toll. While the bulk of those problems stem from the fact that the public Internet is a shared medium, Layer 3 networks are inherently more complex than Layer 2 networks, which saps performance.
• Software Defined WAN (SD-WAN). Option two is typically SD-WAN. Most SD-WAN kits use customer premise devices that are installed on the WAN side of branch office routers and provide interfaces for MPLS and Internet access pipes. You can use policies to dictate what traffic uses what transport, with the hope that mission-critical traffic can be restricted to the MPLS pipe. The trouble is though, the Layer 3 Internet component is subject to the same problems outlined above. As enterprises embrace digital, more applications are considered mission-critical – the data no less important, so this option fails on two fronts.
• Layer 2 private SD-WAN delivered as a service. SD-WAN becomes more compelling if you can take the public Internet out of the equation, especially for the middle mile where roughly 95% of the network bottlenecks occur. Replacing the Internet with an SD-WAN as a Service delivered over a managed, optimized, high performance, Layer 2 private network gives you policy-based controls and a host of other benefits that do it yourself SD-WAN kits cannot deliver.

The fact that you consume the network versus build it is the first big advantage. You don’t need to train staff to become SD-WAN experts. You can deploy in a fraction of the time it would take you to purchase the selected gear and deploy it. What’s more, modifying a service is immensely easier than scaling a network on your own, especially if the SD-WAN is global in nature.

But look for an SD-WAN as a Service that is delivered over an optimized Layer 2 private backbone. Layer 2 networks are simpler and more efficient than Layer 3 networks. They use MAC addressing instead of IP addressing and are switched versus routed, which ensures low latency. That latency, after all, typically results from packets getting buffered at routers in a Layer 3 network.

While some worry about broadcast storms in Layer 2 WANs, that’s the beauty of consuming a Layer 2 SD-WAN as a Service versus a DIY approach. The service provider is the expert at running the network and the enterprise simply benefits from using a fast, flat network mesh.

Obviously, you’ll want that Layer 2 network deployed over a redundant mix of Tier 1 facilities, such as DWDM or Ethernet-over-SONET, and not over MPLS. And you’ll want to ensure the service provider has points of presence (POPs) close to your facilities and users.  In fact, you’ll want to ascertain how many milliseconds away most of your resources are from the POPs so you can ensure the new SD-WAN will be able to deliver reliable and consistent application performance.

Also key to application performance is what the service provider does within the core to optimize traffic. If they get the Layer 2 foundation element right, they will have limited packet loss and stabilized latency, which should mean they can game TCP to limit packet acknowledgments, which will result in huge throughput gains. Which is to say, an SD-WAN delivered over a Layer 2 private backbone is the best way to deliver consistently high application performance, regardless if your users are across town or on the other side of the world.

And finally, make sure the Layer 2 SD-WAN service provider has off-ramps to prominent cloud service providers so that creating links to those resources — regardless of where they are homed – is as easy as spinning up the cloud resources themselves.

There are many WAN options that can help you survive the perfect storm tossing around IT today, but when you’re evaluating SD-WAN solutions, carefully determine if the core network will be an enabler or an inhibitor for your efforts going forward.

Want to learn more? Download our white paper, SD-WAN vs. MPLS: Key Considerations for Your Global Enterprise Network.

About the author

Andy Leong, Sr. Director, Product Marketing

Andy is the Senior Director of Product Marketing at Aryaka. He is responsible for the company's go-to-market strategy, outbound marketing, and corporate communications, bringing over 20 years of leadership experience from across networking and software industries.