5 Reasons Why We May See the End of the Private Network
There's been plenty of hype on the topic of the cloud, and whether the cloud could eliminate private IT. Not much is being said about a more pressing question, which is whether we're going to see the end of private networks. Remember the days of T1 lines, DDS, and frame relay? They're pretty much gone. Remember when everybody had their own router and switch networks? They're past tense, in the age of virtual networks. Now we have five new developments that could put the nail in the coffin.
Better Equals Cheaper
The most fundamental of all these developments is that enterprises think the only "better" network is a cheaper one. For the last decade, network operations managers have told me they are under pressure to lower costs, meaning network total cost of ownership (TCO). The easiest way to do that is to use less and less expensive equipment, and to substitute Internet connections for more expensive private LAN/WAN services. All of these steps will reduce a company's commitment to private networking.
But there's more. Corporate influence tends to track to the size of your budget, and network operations managers have lost ground continually. The number of companies that don't even have a separate network operations group, but have lumped it in with IT, has more than doubled in the last decade. Who speaks for private networking anymore?
Networks Served Up as Services
The next critical point is that network-as-a-service concepts are improving and expanding. Virtual private networks were revolutionary because they reduced or eliminated the number of private network nodes required. Fewer nodes meant fewer boxes and less technical support burden. Now we have things like network functions virtualization (NFV), virtual CPE (vCPE), and software-defined WANs (SD WANs) that could reduce or even eliminate private hardware.
vCPE replaces premises appliances with hosted functions, and virtualizing firewalls and VPN services in this manner is already under consideration. SD WANs can use Internet tunnels and private VPN services in combination to serve more sites at a lower cost, and they facilitate a transition to pure Internet VPNs. If you host SD WAN edge elements as vCPE, you end up with no private network devices at all, and enterprise networks could transition entirely to Internet overlays. With no devices or WAN services to manage, what does the network operations group do?
Cloud as Network
Perhaps it collapses into the IT organization. More and more companies consider data center networking to be part of the data center and not the network. If the cloud makes the world into a distributed data center, then what? The evolution of cloud computing is putting more of the network inside the cloud. One of the impacts of cloud computing is the notion of "phantom IT," where line departments buy their own computing services. It's hardly likely that such a roll-your-own IT mindset would then turn to a private network to provide the connectivity. Most of the phantom IT is connected using the Internet or Internet VPNs, not private networks.
Even when there is professional IT development, cloud services are evolving from simple hosting to two dozen or more sets of tools that build unique cloud-specific applications. If all applications migrate to using these special tools, how much network traffic disappears inside the cloud, forever outside the private network domain?
But what about service-level agreements (SLAs) and fault management? Won't a shift to hosted services and devices totally change user management practices, to the point where the transition to the new services would have enormous impact on staff skill requirements? Perhaps not, because software automation of management blurs the public/private boundary.
Both network operators and network users are moving quickly to adopt software tools that manage a service lifecycle. These tools, serving as they do both operators and users and working across a whole spectrum of vendors and technologies, make it easy to change from private networks to public network services and preserve operating practices along the way. The same tools, applied by managed services providers, can support the outsourcing of network operations tasks by users (and even by network operators). If software is really running the networks, it matters little who's running the software.
For most users, that's good news, because network skills are harder than ever to acquire and retain. One CIO I talked with recently quipped that "every cloud startup seems to draw talent right from my own labor pool, even my own staff." Network skills are in demand, and in most industries, it's difficult to pay network professionals enough to keep them from being poached by vendors, network operators, and startups.
SMBs are particularly in a vice here. On one hand, things like online retail, distributed workforces, virtualization, and the cloud are increasing their dependence on network services. On the other, they have no realistic career path to offer for a network professional. Private networking isn't the answer for them even today, and it's less likely to be the answer in the future.
I'm an old-timer in networking, and I have to admit that predicting the demise of private networking makes me uncomfortable. I'm also a realist, and we have a technology history replete with examples of how something that had to become populist had to become very simple. Networking is so critical to doing business today that it can't stay complex and expensive. So, it won't.