As enterprises finalize budgets for 2018, a common question in IT departments is how to budget for the implementation of SD-WAN technology. The fact that these conversations are even happening is noteworthy in itself; this is a technology that has gone from being a curiosity 18 months ago to a top-5 initiative for many IT teams in recent months.\nThere are many variations between SD-WAN vendors and service providers on what the technology offers, how it\u2019s paid for, and how the business case stacks up. Here are a few items to consider:\nFor most enterprises, SD-WAN savings are based on transport savings\nOne of the headline benefits of SD-WAN that attract most enterprises is the prospect of significant savings. This can mean many things, but the bulk of these savings typically comes from replacing private MPLS connectivity with Internet-based services, and using SD-WAN to glue these together. Internet \u201cmay\u201d be poorer quality (more on that later), and may not have performance guarantees, but the path performance tracking and steering capabilities of SD-WAN can work around these limitations. This may be the case, but it isn\u2019t universal:\n\nIt is highly dependent on geography: the MPLS-Internet price gap can be much smaller in North America and Western Europe than it is in Asia Pacific or Latin America.\nYou have to consider broadband to see the biggest savings: replacing a leased line MPLS connection with a fiber-based DIA service often won\u2019t deliver any real savings, as most of the underlying infrastructure is the same. Broadband services, especially asymmetrical products, open the door to far greater savings, but come with some performance limitations.\nSomebody has to manage these all new Internet services: this can be a big surprise for enterprises \u2013 the real savings come from contracting with in-country providers, but this creates a lot of fragmentation. This requires internal staffing or a managed service wrap to deliver a high-quality service.\nYou have to work around your contractual limitations: most MPLS networks are contracted for multiple years, often 3 to 5. A well-negotiated agreement will provide more flexibility, but you may have to start with MPLS in the transport mix, impacting savings opportunities.\n\nYes, there are some big potential opportunities at the transport layer, but it can be one of the hardest areas to get right. Doing a poor job here means you may not find sufficient savings to pay for the new costs of the SD-WAN layer itself.\nFind a commercial model for the SD-WAN layer that works for your business\nIt can be nearly impossible to directly compare SD-WAN offerings, due to the variety of commercial models that are now available. Here are just a few that I\u2019ve seen:\n\nBuy the SD-WAN technology layer as an annual or monthly subscription, including the use of all hardware and software components.\nBuy the SD-WAN appliances and pay annual maintenance and software license fees.\nObtain the SD-WAN technology layer as an add-on to a traditional telco service (e.g., MPLS).\n\nNormalizing the various offers \u2013 somehow \u2013 is essential. A 3-year TCO comparison is a good start; the technology is moving too rapidly to realistically consider longer agreements. This raises an important question about buying hardware: how confident are you that the hardware you buy today will effectively run that vendor\u2019s SD-WAN software in 12 months from now? In 24 months? Subscription-based models, even if they are not the absolute low-cost option, may shield you from the risk of buying hardware with a short life span.\nAll of the usual business-specific considerations on capex vs. opex also apply here \u2013different enterprises have different tolerances and budgets for each type of spend. Depending on the SD-WAN vendor or service provider you work with, there may be some flexibility on how the transaction is structured.\nDon\u2019t lock yourself out of further savings\nTraditional network services were procured on a fairly predictable 3 to 5 year contract with limited ability to track market-level changes in pricing during the term. It\u2019s difficult to overstate how risky that approach is in a new agreement that incorporates SD-WAN. Here are some thoughts on how you can help incorporate the protections that will be needed in this fast-moving industry:\n\nMake sure your transport layer is contracted on a short-term basis. There is continuous innovation in last-mile technologies and competition continues to drive up available bandwidths and keep costs contained. The ability to access these options as they become available will prevent being locked in to high-priced, outdated services.\nLeverage self-service options. The traditional managed service model needs to be refreshed in an SD-WAN environment. Many changes that previously were owned by the MSP can now easily be handled by the enterprise, even with limited staff. Negotiating the ability to perform these self-service changes (and avoid expensive change fees) can help contain costs over the term.\nDon\u2019t automatically reject upgrades. Many traditional networks were deployed with a single tested software image on the routers, and left that way for the term. Monthly code releases are common in an SD-WAN environment, and can bring material improvements in functionality and performance. Negotiating the ability to obtain these updates over the term is essential (including hardware refresh if needed), and making the necessary business process changes to adopt these upgrades should be part of the implementation plan.\n\nEnterprises can derive some great benefits from this technology, but it requires a real plan, especially as it starts to displace ever-increasing portions of existing IT budgets. Recognizing the nuances of each business and avoiding a \u201cone size fits all\u201d approach is one of the most important first steps.