MPLS huggers rejoice!

MPLS huggers rejoice!

In our 2016 Cloud and Networking Benchmark, Nemertes heard a lot of enthusiasm from early adopters of SD-WAN. And a healthy amount of caution.

About 18% of organizations are deploying some flavor of SD-WAN. Mostly, they have been drawn in by twin promises: that SD-WAN will deliver them a better WAN—more reliable, resilient, and performant—and that SD-WAN will also deliver a less expensive WAN (at least on a cost-per-Mbps basis).

The results have been gratifying for them. Most began their deployments with their most troublesome WAN locations: those with the poorest quality connectivity, or the fastest growing demands, or a combination of the two. By adding SD-WAN to these locations and using it to manage not just their existing primary WAN link (usually MPLS) but also an existing or new secondary link (usually Internet) they have seen 95% reduction in WAN service outages and 92% reduction in WAN troubleshooting tickets and time. These reductions in service disruption and IT overhead are happening in spite of, not in the absence of, ongoing problems with individual links. Outright link failures continue to happen, and (more insidiously) “brown outs” during which service is significantly degraded without ever actually failing. Thanks to the SD-WAN solutions they have in place, they tend to know about these problems sooner than they used to, and understand the nature of the problems better (in some cases being able to provide their carrier with important troubleshooting data). But, because of the transparent failover of traffic from bad links to good coupled with prioritization and bandwidth management, link failures have become non-events for end users. When there is no service interruption, IT spends less time (and adrenalin) on resolving individual, usually transient, connectivity problems.

So, promise one is being fulfilled: they are getting that more resilient and more reliable WAN.

As for promise two, cheaper bandwidth, well, they are getting that too, with a significant caveat: they are not getting rid of MPLS. At this point, more than 77% say they have no plans at all to drop MPLS in their SD-WANs (seeing it instead as the backbone of a high-performing SD-WAN), and the rest put their plans to drop it at least 5 years out. However, they are changing their relationship to MPLS; most significantly they are shifting the majority of their short-term growth in capacity to other, cheaper connectivity options.

Clearly, the experiences of early adopters are enough to justify an exploration of SD-WAN for anyone with a WAN. And those who can’t see the logic in dropping or even significantly reducing use of MPLS—or who simply have no confidence in the alternatives on their own—should take heart. They may not be able to reduce current WAN services spending but they can certainly bring down the cost of their future WAN by reducing the cost of growth AND the operational costs of managing the WAN even while improving services to the business.

You can read more about this in our full report, here: and hear more about it in Nemertes’ session at the FutureWAN Virtual Summit (




Share this post