SD-WAN Architecture vs Traditional WAN: What’s the Difference?

As major buzz continues to surround software-defined wide area networking (SD-WAN), those who are interested in the technology but not yet fully on board may wonder what makes it particularly different from traditional WAN architecture — and if it's truly worth the investment of its adoption.

Our latest techtorial will explore the principal difference between the two — namely, their relationship to and compatibility with the cloud — as well as the additional contrasts of these networking options. We'll also take a closer look at use cases in which SD-WAN is clearly the better option than its traditional WAN counterpart.

The WAN Lowdown

WANs, in all their iterations, exist to distribute traffic between sites, internet destinations and the cloud. They collaborate with the local area networks (LANs) that handle telecom on individual organizations' premises to ensure everyone stays connected as needed: remote workers to their company headquarters, business branches to one another and so on.

Despite having been an effective enterprise telecom solution for about two decades, traditional WAN is hamstrung by one particular attribute: its near-total reliance on hardware performance for its success. Everything from the hundreds of feet of cable and dozens of routers in branch offices to the massive servers at data centers is necessary to support a WAN in any form it takes — be it cable or DSL broadband, cellular LTE/5G or multiprotocol label switching (MPLS). This major issue, combined with the number of devices connected to the WAN and the sheer bulk of the traffic traveling back and forth can make things cumbersome for enterprises in most WAN iterations. (MPLS is an exception, but it can be prohibitively costly for some businesses to implement and maintain.)

Beyond all that, WANs are also not truly compatible with the cloud. They're capable of connecting to it, but doing so requires the backhauling of considerable traffic to data centers, causing significant delays in packet transmission and application performance. The larger number of software as a service (SaaS) and infrastructure as a service (IaaS) tools that modern businesses use also exacerbates latency, packet loss and other service issues.

How is SD-WAN Architecture Different from Traditional WAN?

Much like classic WAN systems depend on their hardware, SD-WAN solutions are based in their software. They're virtualized, serving as virtual private network (VPN) overlays atop the same network circuits that would be in any other WAN — broadband, cellular, MPLS and so on. (The SD-WAN solution itself may be either entirely virtualized and software-controlled or deployed in conjunction with specific devices capable of handling the traffic demands. These look like routers, but are typically more powerful than the average router.)

Although SD-WAN is leveraging the same nuts-and-bolts hardware of its non-virtualized counterparts, it surpasses them by aggregating multiple links from different service providers under the same banner and harmonizing the traffic. This eliminates most of the potential for sprawl that can slow down WAN architecture.

What problems does SD-WAN solve?

SD-WAN addresses — and significantly alleviates — problems often experienced by users of traditional WAN systems in the following areas:

• Cloud connectivity: SD-WAN offers a more streamlined and clear path to optimal use of the cloud than traditional WAN by leveraging multiple network circuits. It also allows for more reliable connection to the many Software-as-a-Service and Infrastructure-as-a-Service applications that are critical to modern enterprises.
• Application performance: Through a process commonly called "application steering," SD-WAN identifies the applications an enterprise is using at a given time, classifies them by priority and routes traffic accordingly. High-priority app traffic might go to the best-performing circuit (an MPLS or high-end broadband connection) whereas lower-priority apps may be temporarily diverted to a lower-performing circuit (such as a 5G or LTE backup). Because this happens continuously in real-time, the end-user of the lower-priority app wouldn't notice a significant quality dip, while those using the most important app experience the quality of service (QoS) they need. (Steering also adjusts to compensate for network congestions, brownouts, circuit downtime and other disruptions.)
• Traffic control: SD-WAN offers full oversight and control over traffic in a way that traditional WAN can't match. As such, bandwidth is distributed and used more efficiently. Also, real-time traffic shaping improves operations across the board without infringing on any compliance requirements.

There's also the bottom line to consider: Overlaying SD-WAN atop multiple broadband circuits offers QoS equivalent to MPLS (the highest-quality traditional WAN) at significantly lower cost.

What Types of SD-WAN Deployments Are On the Market?

Currently, the three most common ways in which enterprises deploy SD-WAN are: via the public internet through a broadband connection, as a joint MPLS/internet deployment or through a cloud-native, entirely private but globe-spanning network.

Theoretically, the third of those may offer the best possible performance — but it also undoubtedly comes at the highest cost and least flexible. On the other end of the spectrum, solely internet-based SD-WAN is the most cost-efficient option but can be limited in its global connectivity, which is not ideal for most organizations at the enterprise level. An SD-WAN provider that offers customers the MPLS/internet hybrid, or a similar diversity of access options, may ultimately be the "best of all possible worlds" option.

Enterprises also must choose between a do-it-yourself (DIY) SD-WAN setup and one that's overseen by a managed services provider (MSP). The former will afford the most direct control of the network and initially cost less. However, that edge of affordability and control may diminish over time: Personnel with significant SD-WAN experience must be brought on if an enterprise doesn't already have them (to ensure optimal upfront network design) and right now, IT hiring is very much a seller's market. Additionally, the network infrastructure paid for out of pocket in a DIY SD-WAN will likely need to be expanded upon with time, which can put strain on organizations with OPEX business models.

By contrast, choosing a globally connected MSP for SD-WAN takes a lot of guesswork out of the equation: Enterprises can have all of the bandwidth efficiency, connectivity and application performance advantages they need, while the MSP carefully oversees the SD-WAN solution's implementation and provides support and troubleshooting whenever needed.

Learn more about what GTT can offer to ensure peak SD-WAN connectivity and performance.