Nicholas Carr wrote a revolutionary book nearly two decades ago that foreshadowed precisely where Enterprise IT stood at that time. Let me start by paraphrasing his thesis and then explain why this is so apropos in today’s world.
Carr talks about how businesses used to generate their own power. More than a century ago, industrial revolution companies routinely ran water wheels, steam generators, or coal-fueled power generators to supply the precious electricity they needed. But this turned out to be a sub-optimal solution for a variety of reasons:
- The power was unreliable (it often broke down)
- It was far too expensive to build bespoke power generators
- It lacked agility. If a business needed more power, it might take months or even years to ramp-up
- And finally, it took the business away from what it was really supposed to do (build automobiles, manufacture steel, etc.)
The solution was to buy power from a company formed to focus solely on power generation (i.e., power utilities). The benefits were immediate:
- A much more reliable source of power
- The ability to scale power usage up and down quickly (as required)
- Less expensive
- Allowed the company to focus on their core business
Carr then goes on to point out that IT faces the same precise dilemma: Should they build bespoke computing fabric or consume technology "as-a-Service"? Carr was fundamentally correct, and his observations presaged the great migration to the cloud and SaaS.
Why Not Networking?
But while compute and data have migrated to “as-a-Service,” networking has not. IT still builds bespoke networks – and does so at a great cost! The way we build networks today has the same disadvantages as bespoke power did in the early 1900's:
- Network unreliability (brownouts and outages)
- Too expensive to build bespoke networks
- It lacked agility and security. It takes weeks and months to provision networks
- And, finally, it takes today’s businesses away from what we are really supposed to do
I see the dangers of building bespoke networks every day and believe the answer is clear. Just as Carr told us nearly two decades ago, we need to commoditize networking. Let someone else – a networking utility – build and maintain the network.
What about SD-WAN?
Wait, wasn’t SD-WAN supposed to fix this? Well – no. SD-WAN's goal was primarily to reduce costs by leveraging cheap commodity bandwidth. And secondarily, to make it easier to build a bespoke network. Wrong answer – we don't need an easier way to build the network. We need to STOP building the network.
And then there is SASE. SASE takes us back to the MPLS days with less performance and feature controls within the network (default route control, performance-based routing, and Quality of Service challenges). Plus, SASE has serious network visibility and observability issues, which leads to operational risks. SASE is actually a step backwards.
What Does Consuming Networks from a Network Utility Look Like?
Pretty much like consuming power from a power utility. Provisioning should be clicks and minutes, not weeks and months. And IT concentrates on defining policy, not building the network. And the results of letting someone else build and maintain the network?
- TCO goes way down
- Performance goes way up
- Reliability soars
- Security becomes embedded
- AND … IT can focus on front office, high-value work that transforms the business
Back to the Future
The first power utilities emerged 140 years ago. What happened? Power became more efficient, reliable, and safe. The industry advanced more quickly, free from the need to spend valuable resources generating power. And, advances in power came more quickly (alternating current, solid-state, and a steady progression leading to the all-digital smart grid of today).
We need to do the same for networking. Just imagine what industry will accomplish free from the time-sink of managing their own networking. Nicholas Carr was right.
Michael Elmore is Senior Vice President and Global Chief Information Security Officer of GlaxoSmithKline (GSK).