WAN Virtualization used by Cloud Services Providers, IT outsourcers a boon to enterprises, a threat to telecom SPs
This column is usually targeted primarily towards enterprise WAN Managers and CIOs who are, to use terms Geoffrey Moore defined in his books Crossing the Chasm and Inside the Tornado, early adopters and early pragmatists when it comes to deploying newer IT technologies.
Where the Next-generation Enterprise WAN (NEW) architecture is something that can be deployed now by enterprise, with this post I’m focusing more on visionaries and early adopters among the service provider and cloud services provider crowd, providing a glimpse at what WAN connectivity services might be available soon which go beyond the starkly contrasting alternatives of reliable but very expensive MPLS with strong SLAs, and very cheap ($1.50 – $15 per Mbps per month) but unreliable Internet connections with minimal to no end-to-end SLAs.
RELATED: Why does MPLS cost so much more than Internet connectivity?
WAN Virtualization is the next-generation complement to WAN Optimization
While WAN Virtualization and the NEW architecture offer significant advantages to large enterprise, deployment by large enterprises alone likely won’t make much of a dent in the size of the MPLS market. Globally, MPLS is a more than $13B annual market, estimated to be growing about 7% a year. Enterprises deploying WAN Virtualization to augment or replace their MPLS connections are likely to shave a few percentage points off of that MPLS market growth rate, and might even turn it slightly negative, but alone would be unlikely to shake up the entire business.
While exact figures would be impossible to come by, my belief is that the MPLS leaders in each country (e.g. AT&T and Verizon in the U.S., BT in the U.K.) have made more profits in the enterprise networking business over the last two decades than any company other than Cisco.
As touched upon in my last column, though, Cloud Services Providers (CSPs) deploying WAN Virtualization as a service – call it outsourced WAN Virtualization, or perhaps Virtual Network Operator (VNO) 2.0 – is a currently untapped opportunity that could well be the “black swan” of the MPLS business.
What is a “black swan”? Black swans were presumed by Europeans not to exist, since all swans ever seen throughout history were white, up until the discovery of black swans in Australia in the 17th century. As introduced in Nassim Nicholas Taleb’s book The Black Swan, which I cannot recommend highly enough, a black swan event is one which is an outlier, with extreme impact and with retrospective predictability by many, even though few if any predicted it in advance. Outside of high tech, September 11th is a prototypical black swan. Within high tech, the advent of the personal computer in the 1970s, and the rise of the Internet in the 1990s are the two with which most people are familiar. Within enterprise tech, I’d argue that the rise of Linux at the turn of the millennia, and the overwhelming success of server virtualization in the past decade, thanks to VMware, are two.
RAID for enterprise storage in the late 1980s – storage virtualization 1.0, if you will – is another, if more narrow, one in enterprise infrastructure. No one predicted it in advance. Afterwards it seems completely obvious. It shook up the storage industry and enabled the rise of EMC and NetApp, among others.
WAN Virtualization delivered by SPs of various sorts offers this same black swan possibility. The network service analog to it is the rise of the Internet Service Provider (ISP) business in the 1990s. ISPs were originally built using remote access concentrators for modems and T1 connections using POTS lines and T1/T3 links from the existing telecom monopolies and long distance telecom providers to build their routed backbones.
In the case of WAN Virtualization-based services, service providers leverage broadband ISP links aggregated at colocation facilities to cheap ISP transit using Gigabit Ethernet links within the colo to reliably connect almost any global locations to cloud-based data centers and to each other. [The “almost” refers to the fact that you do need to have at least two different WAN connections available at each location to guarantee reliability. The rise of wireless services makes this likely at all but the most remote business locations.]
As a WAN service replacement for MPLS, such outsourced WAN Virtualization has a number of advantages. Even if the end user broadband links had to be purchased at full retail prices, a markup of 200% leaves room for high service provider margins while still delivering much better price/bit to the end user than MPLS does, given the 30x – 100x price/bit advantage that Internet connectivity has over MPLS today. And while an MPLS provider like AT&T or Verizon will cut their price 50% or 60% on a deal-by-deal basis to keep a customer’s business, they simply won’t cut by 85% – 90%, as doing so would crash their market.
Even better for new service provider entrants, such a solution has very low capital costs. Unlike asset-based carriers, after an initial investment in one or a small handful of colo facilities in the geography on which they intend to focus, service providers only need to deploy capital as they sign up new end-user locations.
The perfect storm here is that the move to public and/or hybrid cloud computing makes such a service more valuable and more necessary for CSPs if they want to move enterprises away from internally deployed IT to cloud-based outsourced IT services. To do this required a WAN access layer that’s reliable, scalable and reasonably priced to access these cloud-based services. While generic public Internet connections easily cross the price hurdle, they simply aren’t reliable enough for most enterprises to bet their business on – which is why MPLS remains the dominant WAN service rather than IPSec VPNs over inexpensive public Internet links. Just as RAID did for cheap PC hard disk technology, WAN Virtualization provides exactly that reliability over multiple Internet connections.
If this black swan is to arrive, from where will the service providers come? This is not an easy question to answer, as there are many possibilities:
- new startups selling to large enterprise enticed by that large, staid high margin MPLS business
- the dedicated Cloud Services Providers, from Amazon Web Services (AWS) in the “middle” to the OpSources focused on large enterprise to local “mom and pop” providers selling to SMBs, all trying to solve their reliability and WAN access scalability chicken-and-egg problem
- folks selling managed network services today, including managed DSL services
- folks selling “traditional” managed services for VPNs or remote monitoring, who may even have started as network equipment VARs
- the Virtual Network Operator 1.0 folks; e.g. the Virtelas of the world
- the WAN Optimization as a Service folks, e.g. Aryaka or the Akamai/Riverbed partnership
- “wholesale” WAN CSPs selling to the various data center / computing-oriented CSPs to allow them to provide reliable and scalable connectivity to their customers; i.e. the equivalent of what LightSquared was trying to do for the 4G cellular business before it hit its technical and political challenges
- vertical managed services folks targeted at areas like contact centers, retail or financial traders
- the big IT outsourcers; e.g. the HP Services and IBM Services folks, especially those biggest in cloud computing
My suspicion, and it is only that, is that the initial winners are likely to be innovators either selling “wholesale” services to the CSPs, or managed service providers with some amount of WAN experience today and good relationships with certain verticals. Longer term, I see those initial leaders plus the HP Services and IBM Services type of folks as the ones who will lead in the disintermediation of the MPLS business and its insupportably high margins. But the low capital costs and high margins possible in this business mean that it is quite possible that it will support a large number of players with varying business models across the managed service spectrum.
Per the classic Innovator’s Dilemma problem, despite their desire to succeed in cloud services, it will almost certainly not be the AT&Ts and Verizons, even if the telecom SPs will likely eventually buy into the business after it’s become significant, just as they did with the ISP business.
Farsighted early adopter and early pragmatist enterprises will benefit greatly from deploying WAN Virtualization and the NEW architecture today, whether replacing or augmenting MPLS, because of the union of lower WAN costs, substantially higher bandwidth and greater network reliability and application performance predictability while rolling out new applications and trying to support cloud computing initiatives. But perhaps the greater industry impact will be when some of the various SP players – and eventually many of them – use WAN Virtualization to disintermediate the telecom SPs and eat from their high margins while simultaneously enabling and accelerating enterprise cloud services deployments and providing customers with a reason to stick with their cloud services.
A twenty-five year data networking veteran, Andy founded Talari Networks, a pioneer in WAN Virtualization technology, and served as its first CEO. Andy is the author of an upcoming book on Next-generation Enterprise WANs.
