If you look at or listen to network operator presentations on next-gen networking, you’re struck by the sense that operators don’t trust vendors any more. They don’t come out and say that, but all the discussions about “open” approaches and “lock-in” demonstrate a healthy disdain for their suppliers’ trustworthiness, and just the fact that major operators (like both Verizon and AT&T in the US) are driving their own busses in their network evolution speaks volumes. What’s going on here, and are operators right in their view that vendors can’t be trusted to serve their interests?
My surveys have shown operator trust in vendors eroded a long time ago. Ten years ago, operators in overwhelming numbers said that their vendors did not understand or support their transformation goals. They were right, of course. The challenge that the evolution of network services has posed for operators is mass consumption, where cost is paramount and where public policy and regulations have constrained the range of operator responses. Vendors simply didn’t want to hear that the network of the future would have to cost a lot less per bit, and that stagnant revenue growth created by market saturation and competition could only result in stagnant capital budgets. Cisco, under Chambers, was famous for their “traffic-is-coming-so-suck-it-up-and-buy-more-routers” story, but every vendor had their variant on that theme.
When NFV was first launched, I was surprised to see that even at one of the first meetings of the ETSI ISG, there was a very direct conflict between operators and vendors. Tellingly, the conflict was over whether the operators could dictate their views in an ETSI group where ETSI rules gave vendors an equal voice. Ultimately the vendors won that argument and the ISG went off in a direction that was far from what major operators wanted.
Vendor domination of the standards processes, generated by the fact that there are more vendors than operators and that vendors are willing to spend more to dominate than operators are, is the proximate cause of the current state of distrust of vendors. Since large operators, the former regulated incumbents, are still carefully watched for signs of anti-trust collusion, operators themselves can’t form bodies to solve their collective problems, and open standards seemed to be the way out. It didn’t work well, and so the next step was to try to drive open-source projects with the same goals. That’s showing signs of issues too.
So far in this discussion, “vendors” means “network vendors” because operators’ concerns about intransigence driven by greed focused obviously on the vendors whose business cases were threatened by stagnant capital spending or a shift in technology. In that same early ISG period, operators were telling me that there were three vendors they really wanted to get strong solutions from—Dell, HP (now HPE) and IBM. Eventually they ended up with issues with all three of these new vendors too, not because they were obstructing the process but because they weren’t perceived by operators as fully supporting it. Neither Dell nor IBM, IMHO, fields a complete transformation solution, and while HPE has such a solution they’ve not completely exploited their own capabilities. Operators, in their view, had no vendors left and no viable paths to standardization or community development. As a last resort you have to do your own job yourself.
If you look at the public comments of AT&T and Verizon as examples, operators are increasingly focusing on self-directed (though not always self-resourced) integration rather than on collective specifications or development. They’re fearful even of what I’d personally see as successful open-source projects like ODL for SDN, but they’re willing to adopt even commercial products as long as they can frame their adoption in an open model that prevents lock-in.
Open models prevent lock-in. Integration links elements into open models. That’s the formula that’s emerging from early examples of operator-driven network evolution. They’re willing to accept even proprietary stuff as an expedient path to deployment but it has to be in an open context, because down the line they intend to create a framework where vendor differentiation can never generate high profit margins for vendors. Their own bits are undifferentiable, and so their bit production must also be based on commodity technology. Open-source software, Open Compute Project servers, white-box switches—these are their building blocks.
So does this mean the End of Vendors as We Know Them? Yes, in a sense, because it means the end of easy differentiation and the end of the notion of camel’s-nose selling, where you have something useful that pulls through a lot of chaff. In a way that’s a good thing because this industry, despite tech’s reputation for being innovative, has let itself stagnate. I was reading a story on the new Cisco organization, and the author clearly believed that one of Cisco’s great achievements was to use MPLS to undermine the incentive for thorough SDN transformation. Not exactly innovation in action, which of course is both one of the operators’ current issues and the path to vendor salvation.
Innovation doesn’t mean just doing something different, it means doing something both different and differentiable in a value or utility sense. If a vendor brought something to the operator transformation table that would truly change the game and improve operator profits, the operators would be happy to adopt it as long as it didn’t carry with it so much proprietary baggage that the net benefit would be zero (or less).
Ironically, the operators may be setting out to prove the truth of the old saw “We have met the enemy and they are us!” Operator vision has been just as lacking as vendor vision, and in many ways lacking the same grounding in market realism versus greed. IoT is the classic example of operator intransigence. They promote a vision of IoT whose problems in an economic, security, and public policy sense are truly insurmountable when at least one alternative vision addresses all the problems.
We have IoT pioneers, ranging from credible full-spectrum giants like GE Global Research’s Predix activity to startup innovator Bright Wolf. Network vendors want in on the game (one of Cisco’s new key groups is focused on IoT) and IT vendors like IBM and HPE have ingredients for the right Big Picture of IoT. I think it may be that IoT will be the catalyst to educate both sides of the vendor/operator face-off. It might also be an indicator that even the current SDN/NFV transformation initiatives of the operators will suffer serious damage from operators’ own shallow thinking. If the transformed network doesn’t promote a vision for what’s likely to be the major new application of network services, it has little hope of helping operator profits.
Because some of the biggest drivers of change, like IoT, are yet to be “service-ized” as they must be, there’s still a chance for vendors to redeem themselves there. The right answer is always valuable, especially if it can help you move on market opportunities faster. But vendors don’t have to wait for these big opportunities; there is still plenty of time to offer a better approach than operators are devising on their own, and at a fair profit margin for vendors. It will take some work, though, and I’m not sure at this point that vendors are willing to invest the effort needed.