Intelligent network services threaten carrier cash cows

Online services have changed the business models and profit margins of application software vendors forced to move from sales and support to services and subscriptions, but the power of the networked operations based upon shared infrastructure to impact all forms of technology business seems without limit.

For years, data services from telcos and network carriers have been an oligopolistic cash cow: a necessary and increasingly important (and sizable) expense for all businesses.

Due to a combination of the sanctioned monopoly of historically regulated carriers and the natural competitive moat resulting from the massive capital required to string last-mile cable, carriers have invariably had an advantage when dealing with any organization needing wide area network connections. In the era of digitization and online commerce, that's every organization.

But the asymmetric nature of the power relationship between buyer and seller is tipping back to towards the customer due to the confluence of inexpensive, ubiquitous broadband connectivity, equally pervasive and cost-effective cloud infrastructure and services and the maturation of software defined network technology that decouples data transportation from higher-level network services.

A collection of network technology, backend services and cheap endpoint hardware, which are collectively known as software defined (SD)-WAN, is siphoning off high-profit enterprise data services like Multiprotocol Label Switching (MPLS) and Carrier Ethernet from traditional telcos. They do this by marrying enterprise-level network features, reliability, security and control with cheaper, faster broadband circuits.

The result is a price-performance combination that neither network engineers nor business executives can resist. It may sound too good to be true, but the conceptual foundation for SD-WAN is fairly straightforward and mature, delivering  VPN-like secure, private network tunnels over public networks.

Although theoretically similar, SD-WAN technology is different to a VPN in that SD-WAN doesn't just provide a point-to-point network tunnel, but a WAN backplane that fully decouples the physical data flow from the logical network via a logically central (but often physically distributed) software controller.

The controller defines connection topology, routing decisions, traffic priority (QoS) and security policy.

Two key elements of SD-WANs are the aforementioned network controller and a CDN-like collection of POPs that are the gateways to a cloud-based backend network.

Superficially, SD-WANs resemble a classic hub-and-spoke network where each remote site connects to a common cloud-based interconnect, the hub.

However unlike a reductionist form of hub-and-spoke network, the simple switch, an SD-WAN cloud is a mesh of very high-bandwidth (multi-Gbps) circuits between colocation facilities and carrier exchange points. Controlling this cloud-based network is software that directs traffic, reroutes around failed connections, logically aggregates physical links, enforces network policy, monitors traffic and provides usage statistics.

SD-WAN makes business sense

The disruptive elegance of SD-WAN is that all the network intelligence resides in the cloud.

Individual connections can be anything that accesses the public Internet: broadband DSL or cable, consumer fiber, legacy T-carrier circuits, wireless LTE, even an MPLS connection.

SD-WAN buyers are now substituting expensive telco services for cheap broadband because even accounting for the cost of SD-WAN service, the savings over MPLS can approach 70%. Lower cost of the circuits is just part of the story, since some SD-WAN services don't require any hardware at the remote site, just the existing cable or DSL modem, while others use low-cost appliances that run $100 or so, i.e. at a fraction the cost of an MPLS router.

Finance is only half the SD-WAN business case since the technology provides much more bang for the buck.

Broadband circuits with speeds of 100Mbps or greater are widely available in the U.S. for a fraction of what businesses pay for slower MPLS or comparable speed Carrier (aka Metro) Ethernet service.

Since SD-WAN can aggregate multiple links, it also makes scaling bandwidth much easier and cheaper merely by adding another circuit, whether from the same provider or, better yet, a separate carrier.

Indeed, due to included services like WAN optimization, extremely remote locations such as construction sites, oil fields or even cruise ships can get acceptable performance using SD-WAN over a combination of satellite and LTE connections.

Businesses leaders have discovered the benefits and are widely deploying SD-WAN. This should be setting off alarms at the legacy carriers. According to a recent IDC survey, 70% of companies expect to have implemented SD-WAN in some capacity within the next 18 months.

IDC also expects the SD-WAN market to pass $6 billion in value by 2020 with a CAGR of 90% over its 5-year forecast. That may be a small fraction of the overall market for data services but SD-WAN is at the early stages of adoption.

Interest in SD-WAN is fueled by explosive growth in the enterprise use of cloud services and the mobility of today's business workforce, with employees using mobile devices to remotely access work applications and data over the Internet via public Wi-Fi and LTE.

Together these trends are straining the capacity of existing WANs and mean that the vast majority of enterprise data traffic will flow to external locales, not within internal data centers.

Cisco estimates that cloud traffic will represent 83% of total data center traffic by 2019, up 21 points in 5 years.

My take

SD-WAN initially appealed to multinationals needing to link far-flung sites served by torpid telecom monopolies that overcharged and under delivered for business-class circuits.

Since it was invariably much easier to procure consumer broadband, what was called 'network-as-a-service' provided a way to securely connect to a private WAN via a cloud service at reasonable speeds and prices without waiting for months.

As the software technology has matured to include a full panoply of enterprise network features with better performance and reliability, SD-WAN has spawned a host of vendors, with various technical merits and business models. In turn, SD-WAN has emerged as a viable alternative for virtually any organization and situation.

As WAN demand continues to grow and take a larger chunk of already stretched IT budgets, businesses facing pressing needs for network expansion should absolutely put one or more SD-WAN vendors on the shortlist for evaluation.

Like any new technology, it's best to start small and with a situation like a new branch office or backup circuits where some glitches are tolerable. However, given that many large early adopters have already ironed out the kinks, the risk-reward is decidedly in the customer's favor. The appeal of trading a multi-thousand dollar a month MPLS circuit for a few hundred on broadband, yet sacrificing nothing in features and security is too compelling.

In both its technology and business model, SD-WAN resembles another disruptive approach to network delivery. SD-WAN allows historically consumer-oriented services like cable companies to compete with legacy network carriers.

Likewise it enables the new class of fiber-first services like Google Fiber and community broadband to leapfrog traditional enterprise carriers because the service business model is so compelling fir the buyer.

The SD-WAN market and its enterprise adoption should be both dynamic and fun to watch over the next few years. Business leaders that ignore the technology are leaving money on the table.