/var/blog
Industry and Technology Insights and reflections by Mike Bushong
/var/blog
What happens to mid-market enterprise networking buyers?
07.11.17

A quick survey of the top technology trends in networking would certainly include things like DevOps, SDN, NFV, overlays (microsegmentation), and now intent-based networking. Each of these has arrived on networking’s main stage to much fanfare about the transformational changes they promise.
 
But have you hugged your midsize enterprise network engineer today? If not, go find her and ask her how much time she has spent today working on these technologies. 
 
Enterprise adoption
Over the years, I have spent a considerable amount of time on the road talking with networking and IT teams of all sizes. I used to ask every room I was in to describe where they are in their own transformational journey.
 
To frame the question, I asked them to imagine a scale from 1 to 10. On the lower end, it’s basically CLI-driven. If you want something, you type something. If you want something else, you type something else. Closer to 10, the world has been fully SDN-ified, totally NVF-ified, and completely DevOps-ificated.  
 
So where are people on this scale?
 
Over quite literally hundreds of meetings, the vast majority of the networking world self-identified as a 2 or 3. To be fair, I did not ask AWS or Netflix (and Google said they were high on the scale). But the sample did include tier-1 carriers, Fortune 100 enterprises, and major Web 2.0 type companies. 
 
I’ll just assert that the loudest voices in the industry (not a vendor comment, by the way) talk as if everyone is crowded to the right, approaching the final ascent from where they are to the full-blown vision of where the industry would be if all of these promises ever materialized in a single network.
 
Fantasy vs reality
For anyone who has been around the networking block a time or two, each new round of vendor proclamations can be a little bit amusing. With virtually every new breakthrough technology, there is someone with scars who remembers the last time the technology made the rounds.
 
Before SDN, centralized policy management solutions existed. We’ve certainly tried automation a few times. And aren’t overlays just tunnels?
 
With any of these announcements, depending on the grey in your hair, you might find yourself skeptically asking why things will work this time. 
 
Innovation basecamps
The real problem we have is that, as an industry, we are no longer iterating around what is actually deployed. 
 
In a normal innovation cycle, you would deploy a technology, and then iterate around that technology for a number of years, bringing new capabilities to what amounts to some baseline solution. This is similar to establishing a base camp in the technology mountains, so to speak, from which you can pursue further exploration. 
 
This is a pretty straightforward way of dealing with new technology. It allows you to essentially buy into a platform, and then use that platform as a means with which to grow. I would argue that this is one of the most important reasons that new solutions need to provide not just new functionality but also a platform for follow-on offerings. 
 
Anything that is not inherently a platform is like setting a flag somewhere on the mountain. You can claim that you have scaled to a certain height, but the fact that you have gotten there doesn’t help you get any higher. Wherever possible, you ought to be setting up a new basecamp, higher than where you were previously.
 
Disaggregation and basecamps
Disaggregation actually supports this view of technology ascension. But it’s not just about separating hardware and software. 
 
By decoupling components, foundational elements can be isolated. You can settle on specific merchant silicon, for example, using it as a common foundation for other elements to be built on top. Fundamentally, this creates an innovation basecamp. Anything that sets a new position from which future innovation can happen is a positive. 
 
These basecamps serve two purposes. As I mentioned previously, they are the platform on top of which other innovation is based. But they also reduce the need to reinvent the wheel, which helps concentrate industry-wide R&D efforts on a smaller set of more useful activities.
 
To some extent, this is why proprietary platforms are not ideal. While they allow a single vendor to land and expand, they don’t benefit the whole of the industry. While most of the discussions about open vs. closed center on lock-in, I think the more insidious effect is fracturing the investments towards a collective future state that ought to be ubiquitously available.
 
What happens when a basecamp is built too high?
Basecamps are powerful when they form a rallying point from which future ascents can start. But if a basecamp is set up beyond where anyone can get, then it’s just a milestone—some future target that might one day be attained. 
 
In the current state of networking, we keep setting up new basecamps. We promise IT teams that they can be more automated or more secure or more software-defined. We create architectures that promise a set of benefits that are obviously attractive.
 
But we then start iterating the technology before anyone gets there. And before the industry has generally reached some level of comfort, we have effectively moved the target yet higher. 
 
Consider the automation movement generally. While we have moved on to DevOps and intent-based management, the vast majority of the people who actually deploy networks still use the CLI as the dominant interface. While we talk about orchestrating intent across heterogeneous environments, enterprises are hiring based on vendor-specific certifications just to make sure people know how to configure the network. 
 
And it’s not just automation. The same is true as we pursue controller-based abstractions, orchestrated virtual environments, and the like. 
 
Back to mid market
So this brings me back to mid-market enterprises. It is true that the top carriers and cloud properties are pouring enough money into their infrastructure teams that they can make progress along multiple fronts. It might be true that large enterprises are managing to make some progress as well.
 
But the mid-market is being left behind. And if the collective industry moves on and then moves on again to the next bigger and more sophisticated solution, at some point, there simply won’t be a future beyond simply deploying the next generation of whoever the incumbent supplier is. 
 
Depending on your position in the whole landscape, this should be either terrifying or exciting. The dominant incumbent certainly benefits from this dynamic, but pretty much everyone else loses. Resellers selling into captive accounts will find that they are easily squeezed to widen margins. And end users who can participate in only a slice of the industry-wide innovation might find that the peloton has caught up and passed the leader, at which point it’s actually even more difficult to move.
 
The bottom line
The mid-market enterprise represents a substantial opportunity. And while a good chunk of that market will inevitably move at least part of their workloads to the cloud, the transition will be difficult. Interestingly, in many ways, the journey more than the destination becomes the meaningful product here. 
 
This means that the most important offerings going forward might very well be the services required to facilitate the adoption of all this new stuff. This creates substantial opportunity for clever VARs looking to move from pushing paper to more consultative selling. 
 
Of course, most people like being places, not getting places. The next few years in networking will be about practitioners, I predict. And that starts where the customers actually are, not where we want them to be.

Top Kudoed Authors
User Kudos Count
4