SD-WANs have changed the way IT departments approach networking—and that’s not an understatement.
In the old days, networking from headquarters to a remote location required manual management. With a software-defined wide area network (SD-WAN), networking engineers can configure and implement an enterprise WAN — based on software-defined networking (SDN) — to more effectively route traffic to branch offices and other locations.
SD-WAN technology is more flexible and agile than traditional methods because it eliminates manual commands, shifting management tasks from hardware to software. That opens a world of possibilities, as network engineers increasingly work in an increasingly multicloud architecture, where infrastructure resides in the cloud, in on-premises data centers and at the edge. This multidomain management needs more flexible software-based approaches, such as SD-WAN.
But, if you’re considering SD-WAN technology, beware some misconceptions that you may encounter. In what follows we'll set the record straight, so you will have accurate information when it comes time to request SD-WAN budget dollars. We’ll cover five myths—positive and negative—that have always been inaccurate or have become inaccurate over time as SD-WAN technology has matured.
Myth #1: SD-WAN provides LAN-like performance in a WAN. While SD-WAN technology can improve the performance of critical applications for users at remote offices, note that the technology is nonetheless bound by the limitations of the WAN circuits across which it operates.
SD-WAN helps improve remote network performance by continuously evaluating throughput capacity, packet loss, latency and other characteristics across two or more circuits. At the same time, it identifies, categorizes and prioritizes data flows traversing the WAN. The most business-critical and latency-sensitive flows are then flagged and sent across the most optimal path at that moment. That said, SD-WAN can’t defy the laws of physics. Your WAN will be only as fast and reliable as the combined circuits you provide for it. SD-WAN doesn't deliver LAN-like capabilities. Instead, think of it as WAN – but with advanced foresight and application performance intelligence built-in.
Myth #2: I have WAN optimization, so I don’t need SD-WAN. Despite the fact that WAN optimization and SD-WAN both focus on solving bandwidth limitations between geographically dispersed sites over leased-line circuits, the two technologies are dissimilar from an execution standpoint. The goal of WAN optimization is to reduce bandwidth demand on a leased line using techniques such as compression, traffic shaping and data deduplication. Thus, more data can be sent over a WAN circuit without creating a bottleneck. SD-WAN, on the other hand, exploits two or more circuits simultaneously while also identifying and prioritizing data to be sent across specific circuits toward their destination.
Depending on the type of data flowing across the WAN, optimization techniques may provide little value. Only certain types of data flows can be compressed without harming the overall performance from a time-sensitivity perspective. WAN optimization also does nothing to address latency and jitter performance issues. Lastly, SD-WAN allows for multiple circuits to be used in an active/active fashion. This creates more bandwidth without compression side effects. The same cannot be said for WAN optimization technologies. Ultimately, most network architects have abandoned WAN optimization in favor of SD-WAN—or they combine the technologies to work together.
Myth #3: SD-WAN eliminates the need for MPLS. I’ve been involved in WAN rollout projects where Muliprotocol Label Switching (MPLS) circuits were indeed replaced with Internet broadband links. However, the end result eliminated the need for only some MPLS circuits—not all of them. Similarly, you may find that your WAN can leverage the power of SD-WAN to remove certain MPLS deployments and replace them with lower-cost broadband alternatives. However, don’t count on it from a budgetary perspective. As remote locations grow and application requirements change, MPLS is still considered the go-to WAN technology for reliable throughput and predictable network latency. Assume that MPLS will be part of your WAN for years to come, even if you choose to deploy SD-WAN.
Myth #4: Implementing SD-WAN guarantees long-term cost savings. Long-term cost savings are guaranteed once you implement SD-WAN. Much of this has to do with the concept of replacing expensive MPLS circuits with lower-cost broadband. Again, this is theoretically possible, but you should never assume it will automatically happen. Instead, think of SD-WAN as an insurance policy against rapid and unexpected leaps in remote office bandwidth demand. Most organizations are discovering that the demand for bandwidth at remote locations is rapidly increasing due to a rise in application usage and data collection. To help offset the high cost of MPLS throughput, increasing secondary broadband circuit speeds can often be used as an alternative – and a price protection – against these types of bandwidth demand spikes.
Myth #5: SD-WAN can’t be used to better manage WAN traffic in multicloud environments
Multicloud architectures are being rapidly adopted and deployed because enterprises want to exploit public cloud providers’ unique services, features and cost savings while bolstering overall application and data access resiliency. Still, multicloud architectures pose challenges, such as how to streamline network and security policy management across disparate clouds. Early multicloud adopters struggled because management platforms had not matured. Another early-adopter struggle was that SD-WAN technologies weren’t widely deployable in many infrastructure as a service and software as a service public clouds. This led many people to believe that SD-WAN cannot provide better WAN traffic management in multicloud environments, but this is no longer the case.
Most enterprise-grade SD-WAN technologies can now be easily implemented inside the most popular cloud service provider networks. Using Internet broadband, dedicated WAN connectivity or centrally located third party point-of-presence (PoP) sites, SD-WAN can operate in and between multiple public clouds as if they were a singularly managed network.
For those that also use popular SaaS applications, some SD-WAN network overlay technologies can be integrated in these fully managed cloud services as well. These platforms commonly bring SD-WAN performance to SaaS applications by calculating data loss and latency statistics to the multiple SaaS cloud entry points into an overall score for each path. SD-WAN artificial intelligence will then analyze the various calculated scores and direct mission-critical and latency-sensitive flows down the most optimal path toward the SaaS cloud.
There are some outdated assumptions surrounding SD-WAN technology that can jeopardize the ROI of SD-WAN in your environment. Be sure to research how SD-WAN can benefit your organization from a performance and ROI perspective. The more planning you undertake now, the more you’ll realize benefits in the future.
Andrew Froelhich is the president of West Gate Networks, an IT consultancy and services provider. He has been involved in enterprise IT for more than 15 years. His primary focus is Cisco wired and wireless, voice-network design, implementation and support as well as network security. Froehlich has experience with network infrastructure upgrades and new buildouts. He's also been heavily involved in data center architectures designed to provide fault-tolerant enterprise applications and services to thousands of users.