Enterprise Wide Area Networks (WANs) have changed significantly over the last decade, moving from basic hub-and-spoke architectures based on services such as leased lines to fully meshed MPLS WANs, and then to hybrid WANs leveraging both the Internet and MPLS. Now we are experiencing a new important shift toward software-defined networking, (often referred to as SD-WAN), centrally controlled overlay networks, which promise to bring simplicity, agility and infrastructure independency in the wide area networks market.
Gartner forecasts that, by the end of 2019, 30% of enterprises will use SD-WAN technology in all their branches, up from less than 1% today. So I would think that of the renewable base, about 50% of all new activations next year will rely on SD-WAN technology and pretty much 100% the year after that. 
IDC estimates that worldwide SD-WAN revenues will exceed $6 billion in 2020 with a compound annual growth rate (CAGR) of more than 90% over the 2015-2020 forecast period. 
Furthermore, in IDC’s recent survey research, 70% of respondents say that they will adopt SD-WAN in the next 18 months (in some form). 
The advantages promised by SD-WAN can be grouped in 3 areas:
1. Simplified WAN management
Zero touch deployment
Centralized controller: Enables enterprises to configure application-based policy and security rules through a single interface, eliminating complexity and error-prone policy changes.
2. Application performance
Dynamic path selection: Measures packet loss, jitter and latency between multiple transport and dynamically rerouting/load-share traffic to the best performing path - obviously the assumption here is that every site is connected to the network via at least 2 different paths.
Visibility of the application performance, which gives a good indication of the actual user experience.
3. Economical bandwidth
Transport agnostic : Leverage broadband internet as transport.
Load-sharing and application performance optimization can reduce the amount of bandwidth required.
Things sound great indeed. But nothing new is perfect.
Now let's see what we regularly hear in the market about SD-WAN and how that ties to real deployment scenarios:
1. The main marketing pitch made by hardware manufacturers such as Cisco, like several sector publications, highlight that SD-WAN will lower costs of the network. In other words SD-WAN saves money over MPLS networks. And this is true if it replaces a traditional WAN. But the cost advantage over a hybrid WAN, which is already heavily exploiting the Internet as transport may not be substantial, or may not be there at all! In some cases the SD-WAN could even bring an additional cost.
So in reality implementing an SD-WAN may turn out to be an expensive project: depending on the hardware vendor, one has to factor in the cost for either high-end devices or additional devices; also every single site should be connected with multiple links or ISPs.
2. Others vendors, such as FatPipe and Network Computing, highlight that SD-WAN can optimize network performance.
To be precise the SD-WAN software could optimize how the applications are transported over the network but does not improve or fix the network infrastructure itself. That will come from a good design and professional management, as it always has; and these are major operational tasks.
And finally some vendors highlight how easy it is to deploy an SD-WAN - claiming that it is "zero touch" deployment.
If we define a service delivery to exclude the delivery and the connectivity and the hardware device(s) on each site, then zero touch deployment maybe “true” for some SD-WAN hardware vendors – but even in such a scenario, this is valid unless something goes wrong! In which case it is typically extremely complex to troubleshoot. Also the underlying infrastructure needs to be deployed as usual.
Transition management can be very complex as well, especially in the case of a phased approach, where the integration or even the coexistence of legacy services with new SD-WAN services may be required - which is the most common use case in our experience.
So what's real?
There are other considerations that enterprise customers should keep in mind: transitioning from a traditional WAN to an SD-WAN can pose serious challenges to the communications especially with solutions that do not allow an easy integration with the existing infrastructure.
Integration and accountability of the network infrastructure is out-of-scope for the SD-WAN vendors, yet even the best SD-WAN software won't solve the limitations of a poorly integrated technology or more simply, it won't solve quality issues associated with the underlying connectivity; so in case of outage, yes there might be a back-up, but for how long can you afford to run on one leg? How much time and resource do you want to invest in managing the infrastructure?
Features that are part of your network services today, such as; secure web gateways, SSL access, collaboration services, private connections, for instance, to Azure and AWS may just not be available anymore on the overlay network and will have to be setup as dedicated solutions.
Connecting to public cloud is not really well addressed by almost any vendor, leaving the solution to the users and their network design.
Security is two-fold: all SD-WAN solutions provide end-to-end encryption; but for those enterprises running on a private network infrastructure exposing the central controller to the Internet might be a serious concern in itself.
Last but not the least, many vendors these days are coming up with an SD-WAN product, and most of these vendors are small startups, and a consolidation in this market can be likely expected: so handing over the lifeline of modern enterprise, namely its network, to a technology vendor that might not be there a couple of years later, for instance to fix a security bug, could be a concern.
Does this mean that SD-WAN is bad for the enterprise? Certainly not! In the right scenario an SD-WAN solution can be of great value to an enterprise customer. I will talk about it more on my next post.
 Gartner, “Market Guide for SD-WAN (ID:G00279081)”, December 2015, by Bjarne Munch, Sorell Slaymaker, Andrew Lerner, Neil Rickard
 IDC Forecasts Strong Growth for Software-Defined WAN As Enterprises Seek to Optimize Their Cloud Strategies (March 24, 2016)
 Source: IDC, “Worldwide SD-WAN Survey Special Report” (May 2016)