Digital Disruption in Service Providers

Every industry can be digitally disrupted - even those that claim to provide the very digital transformation enterprises so desperately seek.

The true benefit of software-defined networking is not separation of control plane and data plane - that doesn't work at scale. It's not easier centralized control and the end of command line with more fancy acronyms like YANG, XML-RPC and REST APIs. The "revolutionary" idea is the separation of the network layer into physical connectivity and logical connectivity. I quote "revolutionary" as tongue-in-cheek - the OSI model always had a separate physical layer and network layer, but in the long move from circuits to fully-managed, bundled MPLS, we seem to have forgotten that.

When all we had was internet, we used it. But then we got private circuits; expensive, but an upgrade. Then we got frame relay; cheaper, better networks without sacrificing security. Then we got MPLS; a win for service providers to make more efficient use of their infrastructure and a win for customers providing a fully managed bundled service at a time when many where cutting costs by outsourcing IT departments thus lacking in-house skills to build and manage networks themselves. Nothing says the next "innovation" can't be going back to the internet - but now with overlay virtual private networks (VPNs).

This is the predicament global service providers find themselves in now, thinking MPLS is the be-all / end-all of networking technologies. More accurate is that any new technology is the penultimate technology - the last ... but one. And when the new one comes along, it too becomes the penultimate in a never ending series of innovations. But innovation isn't linear. Things move in cycles; even bell bottoms made a (brief) comeback (now they're called "bootcut"). And so here we are finding that the next "innovation" is indeed "going back to the internet - but now with overlay VPNs" (now we call it Software-Defined Wide Area Networking - or SD-WAN).

Has anyone learned this lesson yet? The cable companies are learning it now. What a great idea offering broadband over cable television wires and bringing internet to the masses. Better use of their infrastructure, more revenue, job done. But innovation never stops. They might have thought broadband was the end, but cheap, fast connections into everyone's home allowed content providers - not the wires over which content is delivered - to became king. And to be honest, content was always king. Remember "Sneakernet"? Information will find its way from person to person with or without the internet. It may be slower, it may be less efficient, but it will happen. But alas, we have service providers paying lots of money to build the infrastructure over which everyone else can deliver content. In a "virtualized" world, it's becoming less and less desirable to be that service provider.

But what about a different kind of service provider? In a "virtualized" world, anyone can technically be a global service provider. Just resell an SD-WAN vendor's product and provide the configuration service. Host the controller on AWS. Infrastructure light, undercut incumbents' prices, costs are low, margins high = PROFITS! Well ... maybe not that easy, but it certainly means incumbent global service providers should be rethinking their network, cloud and managed service designs. Should you invest in more network hardware if we can now consider network just the routing of information from place to place - decoupled from the physical connectivity? Should you invest in servers for data centers when Amazon has clearly won the market for compute? What do connectivity and cloud services look like in the future when routers and servers are no longer the building blocks that enterprises and end users are playing with?

The new service provider network may look a lot like SD-WAN already discussed. New cloud services may focus more on the content, the data stored and the information that can be extracted rather than the physical infrastructure to host it on. A McKinsey Global Institute report predicts data scientists - people who can expertly analyze and extract insights from data - may be some 250,000 short based on expected demand by 2024.

Hardware is a commodity, there is little but price to differentiate on which only leads to zero. Servers are cheap, but then again so is renting time on them - and it's getting cheaper. Sure you spin up new virtual instances with ease, but then again so can all your competitors. Data scientists aren't cloned so easily. And they are the ones on the leading edge of artificial intelligence and machine learning. They are the ones creating automation that's really making companies more efficient (and "taking our jobs!").

When I first started in information technology, I worked for a consultancy that specialized in network design and routing. All of our customers paid top dollar for talent because they didn't have it in house, they couldn't find the right skills amidst a shortfall during the rush of all enterprises moving everything to the IP-based internet. We used to boast we had more CCIEs than Cisco. Seems like a stable of skilled data scientists consulting out a top dollar may be a good business to start today.