Software Defined Wide Area Networking (SD-WAN) solutions are reaching a crescendo in contemporary Wide Area Network sourcing initiatives. We have not had a single client in the past year that did not at least dip their toe in the water with SD-WAN and in many cases, clients had much more significant plans for adoption. But as the MPLS market had mostly stabilized and most enterprises had to simply consider whether they wanted their MPLS WAN to be with their incumbent or an alternate carrier, SD-WAN disrupted the market, bringing a new technology and new players, and therefore much more complexity, paralyzing many companies with too many options.
Over the last ten years, most enterprise customers focused on just a few requirement areas when sourcing their networks: 1. Who will manage it? 2. What bandwidth and connectivity do I require to each site? 3. What quality of service (QoS) tiers do we need for our applications? 4. What should our Internet ingress/egress/security look like? There are others, of course, but answers to these questions would ultimately form the backbone of the vast majority of requirement building. Thankfully, these questions have been fairly straightforward to objectively answer for most, which translates into a relatively expeditious supplier identification and RFP development process. Unfortunately, however, the introduction of SD-WAN has complicated matters on each of these fronts by adding new, interdependent variables to the equation.
Consider management, the first example. Most companies have a strategy for how they acquire and manage their network equipment and services. Usually, this is something that has been fairly institutionalized and approach does not often change overnight; they either have the infrastructure -CapEx investment, people, processes, etc.- in place or they do not. Do some companies change approach as part of their sourcing endeavors? Of course, but usually this is a cut and dry decision to either make a major shift or not. With the introduction of SD-WAN and its application ranging from provisional, just-in-time access for new sites to primary connectivity for low priority locations, to backup links for high priority locations and the possibility that a hardware manufacturer may provide the technology but the customer will provide the access, suddenly the complexity may easily challenge the most confident self-managed companies. Conversely, the cost and complexity may intimidate the staunchest of companies who prefer to outsource their network management with the feeling of too many moving parts and too much surrender of visibility of control. “So now what? Do we decide to move to fully managed? Self-managed? Hybrid -and which pieces do we take? We don’t even know what to expect and how disparate the pieces will be, so it’s too early to decide so we’ll need more information!”
Bandwidth and connectivity used to be simple decisions as well. Usually, if the bandwidth would remain in NxT1/E1 territory or NxDS3 territory, decisions were made early about incremental increases, since additional loops, router cards, etc. may be required to increase bandwidth. Over the past several years, most organizations have moved on to Ethernet where the limiting factors of bandwidth increases still exist, but are less of a concern due to the broader thresholds for changes to equipment and our local loops. Now that SD-WAN brings a mix of MPLS, broadband, dedicated Internet access (DIA), and 4G LTE, clients have a lot more to consider in terms of bandwidth capacity for each location, and for the network at large, specifically with respect to Internet distribution across the enterprise, SIP, and redundancy. With so many viable permutations, it’s become extremely overwhelming to plan for and even choose which mix is best.
SD-WAN has even thrown a wrench into QoS planning. The ability to balance traffic across multiple links and flood locations with low-cost broadband megabits per second has changed the way many organizations view QoS. Of course, most will stick with a profile that supports real-time applications, but even applications that would typically sit just below real time might be perfectly fine without any QoS applied due to the increased capacity and load balancing made available by SD-WAN. This is dependent on the application, the organization’s tolerance for risk, and, admittedly, whether foregoing the nominal cost (if any) of QoS even makes sense, but it requires careful consideration and QoS profile planning at minimum, nonetheless.
Most enterprises have carefully managed their policies on Internet ingress/egress, typically with distribution based on geography or via carrier interfaces layered into their MPLS cloud. This enables simpler management and tighter security. The introduction of SD-WAN complicates this strategy a bit. First, it theoretically introduces numerous additional points of entry into the enterprise from the outside world. Can these be satisfactorily secured? Depends on who you ask. From a security perspective alone, SD-WAN is not an option for some companies. Second, once again it begs the question of allowing direct access to the outside world at each SD-WAN connected remote node vs. maintaining a centralized model. Navigating the technical and operational pros and cons vs the potential economic benefits is certainly not easy as, once again, it’s technically not a binary decision but rather a mix of options on a per site or per type of site basis.
As SD-WAN continues to rapidly mature and gain adoption, it’s something every enterprise should be considering from a variety of perspectives and the options it brings can bring tremendous operational and financial benefits to the companies who capitalize on them. But the world of options to pick and choose from are daunting, especially coming from a time of much lower complexity from a sourcing perspective. If you’re interested in sourcing your wide area network and taking advantage of SD-WAN and the latest technologies made available from the carriers and hardware manufacturers, contact Source One to learn how to make your next generation network best in class.