How SD-WAN technology is different from WAN

A Wide Area Network (WAN), as a telecommunication network, enables computers, applications, and devices from multiple locations to connect. For example, a company from site A connects to its branch in different geographical areas via a WAN.

Over the years, the WAN technology has evolved to include various aspects like MPLS, leased lines, the Internet, and, more recently, mobile technologies like LTE and 5G. Along the way, other WAN technologies like optimization, security, etc. were added.

The traditional WAN technology was designed for the headquarters centric era when all the traffic from branch locations were routed to a central site. Massive servers for email, web, and applications were located centrally, and all branch locations gained access to them via the datacenter.

Figure 1

When enterprises began to adopt cloud and embark on digital transformation initiatives, the drawbacks of following these approaches became clear.

Here are some of the disadvantages of a traditional WAN:

• Expensive – The solution is costly to scale, mainly when MPLS is used.
• Multiple devices – As seen in figure 1, the traditional solution results in multiple hardware elements, causing a box stacking problem.
• Time to deploy – The lead time for solution deployment is anywhere from 6 months to a year.
• Application QoS – In cases were Internet links are used, the quality of service is an issue.
• Trombone effect – Cloud off-ramps, security, and other vital resources are centralized in the datacenter. The longer path taken by the traffic results in increased delays and performance issues and the loading of expensive WAN links.

SD-WAN – Delinking control and switching

As detailed above, box stacking was one of the traditional WAN problems. Multiple hardware elements were required because standard WAN hardware performs two critical functions – routing decisions and the physical switching of packets. The double duty performed by these hardware elements limits the number of tasks that can be integrated into a single component.

Software-Defined WAN (SD-WAN) solution overcomes this problem by centralizing the control function and leaving the Customer Premise Equipment (CPE) to perform only the packet switching function. This delinking allows the SD-WAN controller to have a global view of the network, resulting in better routing decisions.

An additional benefit of this software-defined approach is the ability to perform regular maintenance activities and software upgrades from a central location with minimal downtime.

Another advantage of integrating multiple transport interfaces like MPLS, 5G, and LTE is that it allows for more intelligent routing of the traffic based on the application’s criticality. For instance, Netflix and other non-business-related traffic can be offloaded to the Internet. Critical traffic can be routed over expensive MPLS links or a private core, thus ensuring the necessary QoS.

Essential elements of SD-WAN solution

Many vendors are offering SD-WAN solutions in the market, and some more advanced than others. Vendors like Aryaka have gone over and above the basic SD WAN solution to provide a Cloud-First WAN as-a-service. But any solution claiming to be an SD-WAN solution must meet the following four essential criteria set by Gartner.

• It must be a lightweight replacement for the standard WAN technology, and it must be transport agnostic with the ability to terminate circuits physically.
• It must use business or application policy to route or load share traffic over various links dynamically.
• The solution must simplify the deployment, management, configuration, and maintenance of the network.
• The solution must provide a secure VPN and have the ability to integrate other network functionality like WAN optimization, security, firewalls, etc. 

SD-WAN solution and the traditional WAN compared 

The following table lists the key differences between the two solutions: