Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

VWAN threatens traditional three-tiered WAN design

With vWAN providers changing the traditional three-legged WAN structure, carriers need to develop new WAN plans and services using SDN and NFV to keep enterprises on their networks.

Recently I wrote about the emerging category of virtual WAN (vWAN) providers that give enterprises options beyond traditional WAN connections. These companies deliver software; hardware; and in some cases, cloud-based services that leverage software-defined networking (SDN), network functions virtualization (NFV) and other technologies (including some proprietary ones) to optimally and securely route traffic across standard Internet connectivity.

In other words, vWANs turn ordinary IP networks into secure, reliable, high-performance WANs at a fraction of the cost of traditional WAN technologies like MPLS.

The emergence of vWAN technologies poses an existential threat to carriers.

The emergence of vWAN technologies poses an existential threat to carriers.

Obviously, physical infrastructure continues to be required, as do layer 2 switching and layer 3 routing. But high-margin services like Multiprotocol Label Switching (MPLS) are increasingly at risk because users can obtain the reliability, security and performance from the vWAN provider rather than from the carrier. Using a vWAN provider, the only service users would need to purchase from the carrier would be low-cost, low-margin basic Internet.

Shifts in traditional WAN structure

I'm obviously not saying that vWANs will wipe out traditional WAN services tomorrow. But to understand the trends at play, it makes sense to revisit shifts in WAN architecture over the past 15 to 20 years.

The image to keep in mind is what I call the three-tiered WAN, which is the typical architecture most enterprise customers rely on. The key point is that the three-tiered WAN comprises different technologies for different connectivity requirements:

  • Data-center-to-data-center, or tier one, connectivity, requires high-bandwidth, low-latency services. Network complexity is limited because the typical enterprise has half-a-dozen data centers or fewer (even including connections to public cloud services like Amazon Web Services, for example). As a result, there's little to no need for routing. In previous decades, tier-one connectivity used synchronous optical network, or SONET. These days, carrier Ethernet is used increasingly.
  • Branch-to-branch, or tier two, connectivity requires medium-bandwidth, medium-latency connectivity. In sharp contrast to data-center-to-data-center connectivity, however, the branch-to-branch-topology is complex. If a company has 1,000 branch sites (not unlikely for a midsize retail chain, for example), the number of interconnections is nearly 10 million (1,000 times 9,999, or 9,999,999). This means routing is virtually required, given that one of the core functions of a router is to make this type of complexity manageable. In previous years, this tier used frame relay. MPLS began replacing that in the early 2000s and is fairly standard today.
  • Remote-site and small-branch, or tier three, connectivity is largely provided by Internet services. Although the vast majority of corporate WANs historically comprised branch-to-branch and data center connectivity, some sites are almost always out of reach of traditional WAN services. As far back as the 1990s, this tier of the WAN has been served by Internet services.

Nemertes Research has documented this three-tier architecture in multiple research projects over the years, and we've noted two accelerating trends.

  • The first is that the amount of traffic traveling over tier two, the MPLS portion of the WAN, is decreasing as enterprises increasingly locate resources off-premises (most likely in the form of SaaS, but increasingly as IaaS and PaaS cloud resources).
  • The second is that the increased use of the Internet as a WAN, typically found in smaller, tier-three sites, is slowly encroaching into more traditional tier-two sites as well. In a small but growing handful of companies, Internet services have become the de facto tier-two choice, replacing MPLS entirely. In fact, the use of MPLS services overall peaked about 2012, with Carrier Ethernet (the predominant tier-one WAN choice) overtaking MPLS in 2014. Just 68% of Nemertes' benchmark participants reported using MPLS in any capacity, while 83% use Carrier Ethernet.

VWAN trend accelerates

Those changes bring us back full-circle to vWANs. Replacing MPLS with Internet services is already a documented trend, and vWANS are accelerating that trend by making Internet services more secure and reliable than ever before.

In the process, they're pushing carriers into the role of commodity providers of basic Internet services -- unless the carriers take action. But what action should that be?

Ideally, carriers are looking at vWAN providers and leveraging these technologies to deliver high-value WAN services across their lower-cost, lower-margin IP services. Developing all new cost and support structures may take some work, but if the alternative is to become obsolete, carriers would do well to sit up and pay attention.

Bottom line: Carriers need to care about vWANs. If they haven't started caring yet, that should be a resolution for 2015.

Next Steps

When enterprises should choose Carrier Ethernet or MPLS for the WAN

Using SDN to make the WAN more functional, less expensive

Listen to the podcast: Dos and don'ts for implementing SDN in the WAN

SDN helps meet cloud WAN demands for programmability

This was last published in January 2015

Dig Deeper on Telecommunication networking