Things are happening in data center networking. I’m not referring to some new technology or visionary trend. I’m referring to an immense structural changes in the way networking technology is consumed and the effect that has on both operators and suppliers.
We all feel the effect of these changes. Ethan Banks and Greg Farro from PacketPushers often discuss publicly what’s to become of network engineers and coach them on maintaining job relevance. Powerful network technology suppliers are changing focus toward innovation in adjacent technology areas. And as we all know, many hyperscale public clouds, the world’s largest consumers of networks, have long abandoned commercial networking hardware and software. And now they’re on the way to the enterprise premises as they push their public cloud offerings directly into private infrastructure (AWS Outposts, Microsoft AzureStack).
What we know for certain is that it’s really hard for enterprises to build and run data center infrastructure (including networks) and doing so may just be a really bad business decision. Does any sane person really want to repetitively design and assemble the guts of a data center, the most complicated system ever devised, all by themselves forever? The market, over time, tends to destroy conditions like this. And the market has never been wrong about anything over the long term.
Public cloud is the first manifestation of the industry’s rejection of crazy complexity. But not everyone will flee there for a zillion reasons we all understand. The market’s answer to crazy complexity appears to be converging on three choices for enterprises, so take your pick;
- If you don’t go to public cloud, then public cloud will come to you.
- If it’s too complex for you, punt to an MSP and let them take the stress and the blame.
- Deploy new embedded systems that just work.
The last item is interesting. Enter Nutanix as an example. They’ve proven something in the market we technogeeks tend to overlook. A meaningful single-pane-of-glass over compute, storage, network, security, public and private cloud, hybrid cloud, multi-cloud is a profoundly awesome thing. They call this what it really is: Invisible Infrastructure. (Network engineers: Invisible infrastructure means invisible to the Nutanix operators, not network engineers who fix stuff. Intent-based systems know precisely what’s wrong and can point network engineers to the right place. It provides them the ultimate in visibility.)
This market obviously likes Nutanix’s idea. But, what does that say about networking exiting as a data center silo and being relegated to being just another embedded system within invisible infrastructure? Well, embedded systems aren’t noticeable at all unless they’re broken. Embedded really means forgotten. It also means that people probably won’t operate them directly. Robots are much better at that. An intent-based network understands how all adjacent embedded systems work, and they can react to changes within other embedded systems instantly. People generally suck at that.
Replaceable components within an embedded network system are just parts with little magic. Pick whatever switches you want and let the system configure, deploy and monitor them. Change your mind and change components like you’re changing tires on a car. Component differentiation becomes unimportant except to the designer of the entire solution and the person with the purse. Eventually, commoditization changes every industry, even networking.
With a deep understanding of this market transformation, Apstra has entered into a powerful partnership with Nutanix (Nutanix Ready; AHV Integrated) to extend Nutanix’s concept of invisible infrastructure to intent-based networks composed of parts from any network supplier. This capability is in trials with major enterprises today. We’ll announce it formally in a little while, but if you’re interested in a sneak preview, please stop by our booth B-09 at .NEXT in Copenhagen this week.