It used to be that we’d look at tech purchases in terms of what buyers like. We may now be entering a period when what matters is what they hate. A big part of this attitude adjustment relates to the shift of tech purchases from new projects to maintenance budgets, and another part is due to increased difficulty buyers of tech experience in getting and retaining skilled workers. Let’s look at how this shift is impacting both enterprises and network operators.
When you go back a couple of decades, you find that buyers’ most-expressed concern was “vendor lock-in”. They saw their vendors’ strategy of building one of those “circular pull-through” ecosystems, where if you bought anything from them, that thing made you dependent on a whole family of products that were associated with your purchase, and which then let the vendor mercilessly exploit you by charging more and more as you got less and less mobility of vendor/product choice. Or they saw a vendor abandoning something or going out of business, leaving you high and dry.
Today, things are different. Today, what buyers fear most is…integration. The whole that’s made up of the sum of parts is a whole that needs assembly instructions, and that’s bad. Then, when the whole breaks, all the parts will point to each other and shrill out accusations. That’s bad, too. Finally, every player in the integrated whole will gradually evolve, so the whole can’t be assembled from the original parts any longer. That’s fatal, because you now have to start over.
The assembly issue is the source of the classic objections to integration. To get X numbers of vendors’ stuff stuck into a cohesive system, the buyer has to be willing to take the lead, or has to employ an integrator. The former requires a significant level of tech skill, more than enterprises think they can easily acquire and retain. The latter adds costs, and increasingly enterprises are reporting that the integrator is just “another finger in the circular pointing”. In addition, the percentage of enterprises who say they can trust an integrator has fallen from 69% in 2010 to 33% in 2024.
The question of whether integrators are just more fingers in the pointing contest has also seen a view shift. In 2010, only 11% of enterprises cited issues with integrator finger-pointing, where in 2024 it rose to 77%. Most of this was due to a radical jump in the number of situations where integrators failed to meet their commitments to get things working together. In almost half these cases, the integrator blamed the enterprise for failing to address all the things expected of the integrator in the negotiations and contract, and about a third came from cases where the integrator blamed one (12%) or more (20%) vendors, but failed to “break the tie” between the parties.
Finger-pointing in the operations period is at least consistent; just over 80% of enterprises reported it in 2019 and 84% in 2024. I’ve been called in to address the fall-out of this particular problem, and it’s incredibly complex and difficult for enterprises to even identify someone qualified to do the job. What’s needed is a pretty deep understanding of the specific issue involved, which of course is difficult if that issue hasn’t been identified. I have to admit that all my experiences here came about because I had a preexisting project relationship with the enterprise involved, so they knew me and my areas of expertise.
The final problem, that of divergent evolution, has grown from only 7% of enterprises reporting it in 2010 to 34% reporting it in 2023. What’s especially scary about this is that the cause of the problem is linked to the open/open-source movement. Fairly, it should be linked to any sort of standardization.
One early text on computer networks comments “The wonderful thing about standards is that there are so many to choose from.” Very true, and even so with open-source software. Let’s turn to the operators for an example. Back when Open RAN was planned, almost 90% of operators said it would assure interoperability of elements. In 2024, when it was being deployed, only 40% of operators said that it had done so. This, in my view, why Light Reading’s article recently suggests that operators are blowing kisses at open ecosystems while buying from the same few giants.
I think that all classes of buyers are now seeing “open” or “standard” approaches as “catastrophe” insurance not as a general coverage contract. If my chosen vendor is open and fails, worst-case, I can hope to juggle in a replacement without a crushing integration cost. Lock-in? So what? Finger-pointing is eliminated with only one finger in the game, and if that player goes away I have an exit strategy. Ugly, perhaps, but if I pick a big enough vendor my first time, the risk of their failing is limited.
This seems to me to be the challenge of open-model networking when applied to building a cooperative system, rather than a set of discrete devices with standard interfaces. Enterprises are, today, less likely to believe that even routers will interwork seamlessly across vendor lines. In 2010, 13% said they could not be relied upon to do that, while today 37% have that view. But expand to the level of a system of devices, something like Open RAN? Enterprises don’t have direct experience with that at levels I can say are statistically significant, but only 14% of enterprises say they’d get something like Open RAN from multiple vendors.
A lot of the old factors that drive vendor and product selection are falling by the wayside, but in an uneven way. Enterprises, for example, love open-model data centers based on Linux, standardized hardware, and even standardized middleware, but they’d really like to get it all from some source who can put it together. The same is true for operators and for networks. It’s great to have open technology, but it’s essential to avoid integration issues. These changes are driven by shifts in the nature and extent of our technology commitments, and they’re going to keep coming, keep influencing buyers and sellers, for years to come.
