Software-defined WAN has many upsides for the typical organization with a WAN of sufficient size. The technology...
can reduce the amount of staff time required to manage the WAN, reduce WAN and site downtime, improve application performance and dramatically reduce costs incurred from increased WAN bandwidth. But IT can't just drop any product into a network infrastructure upgrade and expect it to succeed.
SD-WAN adoption is spreading surprisingly quickly for such a young technology. But in ongoing studies of WAN economics and SD-WAN, Nemertes Research has seen a familiar pattern: IT deploys SD-WAN when it comes time for a network infrastructure upgrade.
Sometimes, refresh is driven by the age of network components or by contract lifecycle. But when the time comes for a network infrastructure upgrade, IT has to decide whether to replace with do-it-yourself infrastructure, a traditional managed WAN or network as a service. No matter which way IT goes, SD-WAN will be central to nearly all plans.
Consider the branch stack when choosing DIY
If IT chooses to go the DIY route, it then has to decide whether to stick with its existing branch stack at each location. This stack usually is made up of three or four devices: the router and some mix of firewall, optimization and wireless LAN controller.
The branch stack may have design principles or security policies that dictate a separation of duties. If those policies can be dropped or relaxed, or if the stack grew over the years without such requirements, then SD-WAN and virtual customer premises equipment platforms allow IT teams the option of collapsing the stack onto a single device -- or, at most, a pair of devices in a failover or hot-hot configuration.
Managed SD-WAN benefits both enterprises and providers
If IT decides to hand off WAN management, substantially change how its existing managed WAN is delivered or look for a new provider, managed SD-WAN will likely be its No. 1 option. After all, SD-WAN offers enterprises new possibilities for connection paths and technology diversity, cheaper bandwidth and improved WAN performance.
On the flip side, SD-WAN offers managed SD-WAN providers lower marginal costs for managing customer WANs. This is, in large part, thanks to SD-WAN's centralized, policy-based management architecture and to the fact that service interruptions on a given link become less urgent to resolve when multiple paths are available and services can continue uninterrupted.
Network infrastructure upgrade considerations
Some of the biggest considerations for an enterprise during a network infrastructure upgrade revolve around available connectivity for each specific location. Companies want SD-WAN to incorporate lower-cost internet connectivity as a primary WAN transport, even though most that have MPLS or Carrier Ethernet plan to retain it within SD-WAN. SD-WAN adopters also want the ability to have diverse connectivity providers in order to improve resilience and stabilize performance.
In some locations, business internet services may not be significantly cheaper than MPLS, however, and consumer broadband might not perform well enough or have enough support to be viable options. IT teams might not be able to achieve path diversity, as there might be only one physical route to a building, for example. Additionally, there could be only one provider in an area, making it difficult to achieve provider diversity. This is often a significant concern for companies with locations in remote or rural locations, such as manufacturers.
Some enterprises plan to keep management of last-mile relationships -- and the need to deal with all these issues -- in-house. Others plan to outsource it through a managed service provider or connectivity aggregator, even if they intend to deploy SD-WAN in-house.
When incorporating existing connectivity into the transport mix, IT also has to deal with the question of the medium. If the existing link isn't handed off as Ethernet, they will need to adapt it for SD-WAN services that don't accept other connection types -- which are most of them. Some enterprises retain their routers in this capacity, essentially using them as media converters; others look for more compact and energy-efficient boxes to drop in line. Others will instead require an SD-WAN option that can accommodate the older link interfaces, of which there are a few.
As it is with connectivity, so it is with protocols. An SD-WAN appliance may or may not speak all the protocols an organization requires in a branch box. The overall architecture will determine which protocols the appliance needs to understand. If the SD-WAN appliance has to interoperate with conventional routers, for example, it should speak Border Gateway Protocol (BGP). SD-WAN products meant to replace routers generally support current open routing standards like BGP. But if an organization's network has vendor-specific or older standards in place, it will need to carefully evaluate an SD-WAN product's ability to interoperate.
IT shops on the cusp of a network infrastructure upgrade should plan for SD-WAN deployment -- if not now, then the next time around. Either way, they should be thinking ahead to the collapse of branch stacks, the primacy of Ethernet as a link handoff and the need to get rid of any lingering remnants of routing protocols past. They should also be actively reviewing connectivity options in all their locations or finding someone -- a managed service provider or a connectivity aggregator -- who can do that on their behalf.