SD-WANs were supposed to reduce WAN costs. After all, Internet bandwidth can be as much as 90 percent less than MPLS bandwidth. It would make sense to assume that SD-WANs would show similar cost savings, but do they really? Not necessarily and here’s why.
Back in September, Steve Garson at SD-WAN-Experts and I analyzed the pricing of SD-WANs and MPLS for one of Steve’s customers. The company was spending $15,200 a month on MPLS and wanted to reduce his spend on his 11- node US network of one data center and 10 branch offices. You can read the details of the analysis on Steve’s blog, but the bottom line was that the customer would see savings of 24 percent, significant but hardly 90 percent.
The exercise underscored one major reason why SD-WANs aren’t as cost effective you might like to believe. As we noted in the blog, SD-WAN providers like to assume that low-cost Internet access is available to every office anywhere, but often that’s not the case. Steve’s customers needed to pull cable to the customer premises, increasing DIA costs by nearly six-times.
Since writing the blog, Steve and I touched on two other factors that can increase your SD-WANs cost. The first is the operational overhead of managing the multiple ISPs needed to build-out an SD-WAN. Good SD-WAN design practices call for multi-homing locations. By connecting offices to multiple ISPs, organizations achieve two goals. They improve availability by being able to switch to a secondary connection in the event of an outage. They also improve application performance as the SD-WAN devices can select from more paths to find one that best matches application requirements.
Managing providers, though, adds operational complexity. As Steve and I explain in our Network World blog, you now have to provide the management that was previously handled by your MPLS providers. Vendor selection, contract negotiation, billing, emergency services and more all of those responsibilities shift back to the enterprise. Global Managed Internet Providers, also called Managed Network Operators (MNOs) and Virtual Network Operators (VNOs) provide one approach to addressing this problem. See this post and checklist for evaluating MNOs/VNOs.
A second factor, are the additional security costs of SD-WANs. If organizations are to take full advantage of SD-WANs, they need to add Internet access points throughout their organization. In the most extreme case, every office participating in the SD-WAN requires a direct Internet connection. But even where organizations establish SD-WAN between regional Internet access points, additional firewalls, URL filtering, malware detection and more needed to secure those Internet points.
I’m still bullish on SD-WANs, don’t get me wrong. But as PWC discovered in a users survey, costs savings are typically a driver for prospects considering SD-WANs whereas agility benefits are prioritized for those who’ve actually deployed SD-WANs. The flexibility and agility SD-WANs engender has been a long time coming. IT can finally match application requirements to the network and, yes, that will often lead to cost savings. But let’s not fool ourselves 90 percent is hardly a fair expectation for what you’ll save in your deployment.