If current market predictions are any guide, SD-WAN is about to take a big bite out of dedicated MPLS services in support of application workloads in the cloud.
But just because it’s new, does this make SD-WAN necessarily better? And if it is better overall, are there any drawbacks that IT needs to be aware of before it throws the switch?
According to IHS Markit, nearly three-quarters of mid-sized to large enterprises are conducting SD-WAN trials at the moment, with a large portion expecting to transition to live production later this year. The need is clear: with application services becoming increasingly important to the business model, organizations need a low-cost yet flexible way to move data over the wide area. Once the floodgates are open, expect the rise of SD-WAN to be rapid, perhaps to the tune of 20 percent growth per year for data center traffic and 30 percent for branch offices and remote sites.
However, the thing about SD-WAN is that underneath all the abstraction it still rides on the same choppy, unreliable Internet that gives us fits when we are trying to buy shoes online. ThousandEyes’ Angelique Medina notes that while SD-WAN gives you better traffic visibility and performance metrics than MPLS, it still requires a shared ISP backbone and all the complex dependencies and best-effort service capabilities of today’s unpredictable Internet. So even if you build your SD-WAN on a dedicated circuit with top-notch SLAs, don’t expect it to be a magic wand for availability, performance and other issues.
Still, it’s a pretty safe bet that many problems encountered on the SD-WAN can be handled fairly quickly. Already, we are seeing the first iterations of artificially intelligent WAN platforms that promise things like global fabric-style connectivity, continuous monitoring and proactive traffic management. The Netrolix AI-WAN, for instance, is billed as a fully autonomous wide-area solution that utilizes real-time and historic data to automatically and dynamically adjust network conditions to minimize latency, congestion and other performance limiters. The system supports MPLS, VPLS and VPN architectures for applications ranging from connected offices and retail locations to IoT devices and drones.
But as with most advanced technologies, SD-WAN success largely depends on how they are architected. 128 Technology COO Patrick MeLampy provided eWeek with a list of seven elements to unleash the full potential of SD-WAN, starting with a centralized policy management and orchestration engine. This allows the enterprise to cope with the myriad network boundaries, firewalls, switching paths and other factors that make up a multi-provider environment. Beyond that, the enterprise should build with an eye toward zero-trust security and deployment, service-centric segmentation, open APIs and analytics, and broad application visibility and control. And you’ll likely need the ability to dynamically switch between wired and wireless networks instantaneously without the need for continuous backup tunneling.
Clearly, the SD-WAN would not have made it this far in the test phase if it were unreliable or failed to meet expected performance levels. But it is by no means a panacea for all things networking. As the enterprise gains real-world experience with the technology, it will likely emerge as a critical factor in emerging data environments – offering superior connectivity to many, but not all, enterprise workloads.
Fortunately, this is one area where the enterprise can cross new bridges without burning the ones behind.
Arthur Cole is a freelance journalist with more than 25 years’ experience covering enterprise IT, telecommunications and other high-tech industries.