When assessing SD-WAN benefits, don't ignore SD-WAN costs
SD-WAN offers several benefits, but companies also need to understand SD-WAN costs. Organizations need to consider some critical areas -- and not all of them are obvious.
Software-defined WAN is driving as big a change in WAN architecture and management as server virtualization did in the data center. It will transform IT's relationship to the WAN and the WAN's relationship to the business -- if IT can make the case for deploying it.
SD-WAN benefits -- including reduced WAN connectivity costs for a given level of bandwidth, less downtime, fewer application slowdowns and faster activation of new sites and services -- have been widely touted. But what about the other side of the equation: SD-WAN costs?
Some costs associated with deploying SD-WAN are obvious and will appear in any cost model, including the following:
- endpoints deployed to sites, if any, and whether they are physical or virtual and the hardware on which they run;
- SD-WAN management, which may range from on-premises services with redundant backup in data centers or IaaS all the way to a fully managed service;
- any new network links to SD-WAN sites, plus continuing service on the existing links that are retained;
- any SD-WAN layered services, for example, if a firewall function of the SD-WAN will replace on-premises firewalls; and
- implementation services, which even most DIY folks use because they have such thin staffing on the WAN.
Some SD-WAN costs are not as obvious, including the following:
- upgrades to on-premises network infrastructure and power to handle new or upgraded links;
- upgrades to existing network links to connect them to the SD-WAN, such as providing an RJ-45 connector -- many SD-WAN products will only accept them, and leaving the legacy router in place to function solely as a media converter is not a good option;
- maintaining services on links being decommissioned until after the SD-WAN has been deemed functional and stable for as long as the risk management folks require -- but assume a month at least;
- training for staff, which is often minimal but required;
- recertifications and audits that the new WAN architecture will require to stay in compliance;
- any discounts on other services or gear that may be lost if the SD-WAN is from a new provider;
- managing the business relationship with a new connectivity provider; and
- managing the functional relationship with a new connectivity provider.
These last two points are especially important because taking advantage of the ability to use multiple providers and lots of links at a site leads to situations where a WAN can have more providers than sites. For each provider, the company has to handle the business relationship, including contract, bill review and dispute settlement. Customers also have to handle the technical relationship, including monitoring connectivity performance and managing how service problems will be resolved -- whom to contact, how to contact them and how to track the problem to resolution.
Most organizations do not track the cost of doing business with a new service provider, but it can be estimated based on existing relationships and the amount of staff time required to manage them. These costs can help a company decide whether to use a managed SD-WAN service or DIY. And, if DIY, organizations should adopt certain policies to control vendor sprawl.
A good business case for deploying SD-WAN should include as many of these SD-WAN cost components as possible, and it should cover the entire transition period. For most large organizations, the transition period will be at least 18 months.