MPLS is a successful, well-proven WAN technology. And use of an MPLS network design is not, in itself, the main...
cause of network problems. Thus, it can easily be part of the solution for a software-defined WAN. The idea that there might be a choice to make, that there's some SD-WAN vs. MPLS competition, is wrong.
MPLS is not perfect. IT professionals regularly complain that MPLS is too expensive, slow to provision and not always available where they want it. And yet, they also praise it. For most, an MPLS design provides the most reliable connectivity they have, with 99.99% uptime and great compliance with service-level agreements on low latency, jitter and packet loss. MPLS has rock-solid delivery -- and on this rock, most enterprise WANs are built.
Is an SD-WAN vs. MPLS choice necessary?
SD-WAN has come on the scene, though, with promises of reliable delivery of WAN services across the best-effort connectivity, the internet. Some SD-WAN services basically promise reliability like that of an MPLS network design and performance across business internet and sometimes even across consumer broadband connections. Early adopters are generally reporting good experiences in this regard. There are limits, however; it can be impossible for any pool of connectivity composed solely of internet links to match MPLS network design performance, especially where low latency is required.
Most WAN engineers and managers are risk-averse when it comes to making decisions that can affect IT's ability to deliver business-critical services. They are therefore rationally cautious about scrapping the bedrock on which their service environment is based. Instead, they largely intend to fold MPLS into their SD-WAN plans. For example, according to a recent Nemertes survey, 77.8% of organizations adopting SD-WAN have no plan, at this time, to drop MPLS; 56% don't even plan to make any changes in their MPLS, aside from wrapping the SD-WAN around it.
The logic of augmenting an MPLS network design with SD-WAN is solid both operationally and technologically. Operationally, it reduces enterprise risk to retain the high-performing, proven connectivity at least through the first iteration of the new SD-WAN technology. Technologically, SD-WAN services are generally built to take full advantage of the quality of the links they are provided. If handed a low-latency, low-jitter, low-loss link, SD-WAN can use it as primary connectivity for appropriate traffic classes -- like voice over IP -- even while using buckets of internet bandwidth to handle less demanding traffic.
The real impact of SD-WAN on MPLS
This is not to say the advent of SD-WAN will have no effect on MPLS in the enterprise. After all, 44% of the organizations Nemertes surveyed expect to make changes. Just under a quarter, 22%, said they will, with SD-WAN, cap their MPLS spending. SD-WAN is causing MPLS rates to drop, so some organizations may intend to ride price curves down while capacities increase or while they expand to more sites. About 12% said they will no longer have small sites on their MPLS network at all, while 5% said they would have only their high-value sites on MPLS.
SD-WAN is out there now, raising the threshold for jobs that require MPLS and exerting downward pressure on MPLS pricing as a result. With many enterprises committed to sticking with an MPLS network design even as they deploy SD-WAN for the first time, we can expect to see MPLS remain a significant force in the WAN for many years to come.
SD-WAN's impact on MPLS routing
How SD-WAN could be seen as an MPLS competitor
Are SD-WAN and WAN optimization coming together?