Even though software-defined WAN is a hot and exciting technology, most prospective SD-WAN users don't understand...
how it works and aren't clear why they should buy it. Part of the reason for the confusion is about whether SD-WAN is simply a cost revolution or a true service revolution.
SD-WAN services came about as an evolution to virtual private networks (VPNs). The traditional technology used for VPNs is a combination of IP and MPLS, delivered via carrier Ethernet or other high-speed business access connections. This MPLS-plus-something combination is too expensive for many smaller sites, and it may be completely unavailable in some rural or international locations. As a replacement, SD-WAN can connect public cloud services to a company VPN -- something that's rarely practical using MPLS VPNs. Because businesses want uniform private networking throughout their operations, they wanted a new approach, and SD-WAN provided it.
SD-WAN uses a series of nodes that connect business sites to an SD-WAN VPN by overlaying that VPN on the internet -- or, in theory, any combination of transport networks, including MPLS VPNs. This overlay network works much like many data center software-defined networking (SDN) implementations, which means the connectivity and routing of the SD-WAN VPN can be controlled independently. SD-WAN assigns users in non-MPLS-VPN sites to the SD-WAN VPN, which then shares the address space with the MPLS VPN. They then connect the two VPNs -- SD-WAN and MPLS -- with a hub, and the users all look like a single community.
Cost-based SD-WAN decisions
Because SD-WAN VPNs can use business broadband services that are versions of consumer broadband, the connectivity is a fraction of the cost of MPLS VPNs. This has already encouraged some businesses that use MPLS VPNs to shift some of their smaller MPLS VPN sites to SD-WAN to save money. In theory, if internet connectivity offers businesses enough bandwidth and quality of service (QoS), some companies could shift completely to SD-WAN VPNs and save a bundle of money. That's the cost-driven face of SD-WAN services.
No service provider likes price-based competition, because it leads to commoditization. Competition that takes the form of a race to the bottom tends to create an industry that can't advance due to a lack of return on investment from key players, and network operators are certainly the key players in the networking industry.
SD-WAN services-based decisions
The service side of SD-WAN offers a possible escape from network commoditization, which is why it's important. The SD-WAN service-driven face comes from the fact it can provide two paths to a new service that could earn incremental revenue. The easy path is SD-WAN is already the basis for managed network service offerings that can command up to almost 50% in pricing premium compared with traditional services managed by users themselves. The more complicated path is one that relies on emerging SD-WAN features to improve connection and access management.
Managed services are already a factor in SD-WAN, because the earliest promoters of SD-WAN technology were managed service providers. Simply taking over the management coordination involved in integrating remote internet service providers' access connections with business VPN services is valuable. Managing the relationship with those ISPs if something goes wrong is even more valuable. This is particularly true for remote sites where local technical support -- in-house or third-party -- is difficult to obtain.
Managed services are still primarily a cost play in the minds of many, however, and some operators are concerned that offering managed SD-WAN services might make them appear to be nothing more than another managed service provider (MSP). That's why more network operators are looking at new strategies to offset any revenue losses from MPLS displacement and differentiate themselves from other MSPs.
Operators' managed SD-WAN features
One possible differentiator is the way SD-WAN, as an overlay technology, can mask transitions in operator network technology for the user. For example, SDN can be worked into network infrastructure gradually, under the covers of SD-WAN. Even if users have different service technologies in different geographies, it won't matter as long as availability and QoS are acceptable. Network transformation could be a lot more radical and more successful if buyers of SD-WAN services were insulated from the technology shift.
This still doesn't insulate the operator from price commoditization of SD-WAN services or help them compete with MSPs. For that, operators need visible service features, and the potential for those SD-WAN features is again part of the architecture.
Because SD-WAN controls connectivity, it can apply identity-based policies to prioritize or block traffic, to change transport network options or routing to the extent that capability is exposed, and even to move cloud connectivity around to accommodate redeployment or scaling. That's a benefit in terms of security, compliance and performance that users are probably willing to pay for, as well as a justification to expand SD-WAN beyond simple thin-site connectivity -- which is connectivity to sites that are too small to justify MPLS VPN connections or located in areas where the service isn't available -- to universal connectivity.
Not all SD-WAN service implementations exploit this level of identity-based connection policies and access control. While the number of offerings with this capability has grown significantly in the last year, it's still in the low single digits. Competition among the SD-WAN vendors -- those that provide the software and appliances -- may be the only way SD-WAN feature sets improve radically enough to shift the focus of competition to service features.
Increased SD-WAN service features may be getting a boost, because Cisco said it will provide a way to embed its SD-WAN technology -- acquired from Viptela -- into its router software. This will force every SD-WAN vendor to work harder to overcome the natural positioning advantage this offers Cisco, and feature differentiation is a logical way to do that.
Without the cost savings SD-WAN services can offer smaller sites, it is doubtful SD-WAN could gain any significant market traction at this time. But every successful market fights commoditization with feature differentiation. In the long term, the success of SD-WAN will depend on new network-as-a-service features.