What do enterprises, vendors, and service/cloud providers think about cloud computing, networking, 5G, the edge, and other stuff? How do they see the future of these technologies? I’ve been trying to get those questions answered, and what I’ve found could justify calling the effort a “qualified success”. It tells us a lot about enterprises, but maybe a bit less about the future of key technologies than we’d like.
Let’s start with the cloud, which is surely a big technology revolution that turned into a business revolution. It’s going to continue to fascinate both CIOs and CFOs, but there are already issues that could hurt cloud growth. The biggest one is ignorance of the business case. If we look at businesses (mid-sized to enterprise) considering cloud computing, almost two-thirds think that the cloud is inherently cheaper than the data center. In other words, move to the cloud and you save money, period. This is absolutely not true, and it poses a threat because failures of cloud projects are already getting a lot of ink.
The reason users believe this fable is that they say cloud software vendors and cloud providers tell them it’s true. Interestingly, when I talk to planners/strategists in both the vendor and cloud provider spaces, they say the biggest cloud benefit is scalability and resilience and that it is not “typically” a good idea to just move an application from the data center to the cloud and expect to save money. So why are their sales people pushing the fable? Because it’s easier to sell that way, and because most users who undertake a cloud project that doesn’t prove out financially (54%) will stay with it despite the flaws.
How about networking? We’re always hearing about this or that transformation in network technology, from white boxes to network-as-a-service. That’s interesting because 69% of enterprises say that the are not planning for any transformation in their networks. Yes, almost three-quarters say that they believe that their approach will shift over the next five years, but they’re not expecting it within their current planning horizon, which is on the average 17 months.
The thing that they believe might be transformational for them down the line is the “white box”, the commodity hardware switch that’s combined with a network operating system (NOS) to create a switch or router. Almost all the enterprises that think things will be different in five years think white boxes are the reason, yet only 11% say they are actively deploying them in anything but specialized applications or pilot tests. The reason is that enterprises believe that evolving from their current network commitment would be difficult, both because of integration and because of write-down issues.
Integration problems, say enterprises, are really finger-pointing problems. When a white box is introduced, incumbent vendors will blame the new device for any subsequent network issues, and enterprises said that white-box vendors were less able to counter this than proprietary competitors to their primary vendor. Thus, enterprises often had to do a lot of the heavy lifting.
The write-down issue is the biggest problem. Any time a capital purchase is made, it’s depreciated over a period of years, as set by tax codes. If the product is displaced, the company has to take a write-down of residual depreciation, and (usually) the cost of the write-down is included in the cost of the project. That makes it hard to displace anything not fully depreciated, but in particular things recently acquired. Since networks are regularly modernized, all this means that it’s hard to make extensive changes to an in-place network. That tends to focus white-box modernization on greenfield areas, or where other factors dictate a major modernization.
Vendors universally downplay or downright ignore these issues, which of course doesn’t make them go away. The typical vendor strategy is to identify a “sponsor” within a company, provide them with information, and through them influence the company to at least initiate a project. Five vendors told me that their approach was not to talk about phasing in white boxes for financial reasons, but to talk about “easing in” or “getting experience”. They’d suggest areas where there was only limited current technology to displace or integrate with. They also said that once a project got started, it was often possible to keep it going.
NaaS is perhaps the most perplexing of the network shifts enterprises talk about. Almost 80% of enterprises say they think that NaaS is “the way of the future”, but even enterprises who have adopted what could be seen as on-ramp technologies for NaaS (SD-WAN and SASE) don’t cite specific NaaS plans. Why? Because nobody seems to be offering them the service.
According to both SD-WAN and SASE vendors themselves, selling their products by making a connection to NaaS is not a current priority. These vendors see NaaS as a carrier offering, and see users getting their devices through the NaaS provider. All of the vendors are interested in selling NaaS to the carriers, but they’re not pushing it at the enterprise level, and carriers are not yet committed to NaaS so they’re not pushing it either. Thus, NaaS could likely happen if there was a wave of NaaS service offerings and promotion by carriers, but not as a spontaneous enterprise requirement.
Then there’s 5G. Not surprisingly, 5G is the new technology that’s generated the most hype and most confusion in the market. Two specific areas where 5G hype and confusion are rampant are private 5G and new applications that depend on 5G services. Enterprise views here differ sharply from both the view of vendors/operators and the media coverage.
Only 19% of enterprises actually think they might be interested in private 5G, though 31% say they’ve been promoted for the concept. When I ask vendors who offer private 5G, they say, on the average, that “over half” of enterprises are actually planning for the service (only 5% of enterprises said that to me). The problem here, according to enterprises, is that it’s not clear to them what applications would justify 5G, making that second area of confusion perhaps the most critical.
Enterprise views on 5G applications tend to divide according to the extent to which expert planners have assessed the situation. Where no expert assessments have been made, enterprises are now expecting cloud providers to offer them those 5G applications, out of relationships now growing between 5G operators and the cloud providers. Since they say that’s not happening yet, there’s no sense of planning because there’s nothing to plan for. Among those enterprises who have had expert planners look at 5G, the almost-universal view is that they’re not sure what a “5G application” is. What makes 5G different from other forms of wireless, they ask? Latency? Until they see where the applications were hosted (presumably the edge) and understand the cost of the hosting, they can’t make an assessment of value. Interestingly, neither group seemed to think that coming up with 5G-specific applications was their responsibility.
Vendors and 5G operators are firm in their belief in 5G-specific applications, and all vendors and many operators also believe in private 5G. Among the operators, there’s a hope that enterprises who consider private 5G have something specific they’d expect to use it for, and that operators could convert these enterprises to public 5G, perhaps via network slicing. Among vendors, the push is said to come from a need to promote future financial gains, because many 5G vendors aren’t really in the running for telco business.
Is there a common factor at play here? Some thoughtful enterprise planners think there is. They remember the golden age of networking, when companies had information resources but couldn’t get them delivered to the employees, customers, and partners that needed them. Networks then were the limiting factor, say our literati, and now that’s no longer the case. Even 5G and its purported latency benefits isn’t seen by this group as offering a release from a critical constraint. “You can achieve low latency for most latency-sensitive applications through local computing,” one expert told me.
The critical factor now, for all new network and compute technologies, is the right applications. Cloud adoption will increase as we understand, first, what the cloud is really good for, and second, how to write applications that realize those benefits. We’re gaining ground in this area, and so I think the cloud has a bright future.
Changes in networking have to be driven by different, valuable, things we do with them, meaning applications. If no such things emerge, we should expect that network planning will be purely a cost management exercise. The problem now, I think, is that vendors and providers are focusing on a quick sale, and you don’t create new applications quickly. Some hope buyers will figure out those applications, but as my same expert said, “Buyers don’t invent the products”. All the talk about the future of these technologies ignores this need for new applications, and that may actually deter their creation. If the cloud, or NaaS, or 5G are going to naturally explode, why work for it?
They’re only going to explode with the right applications to drive them, and those who promote the new stuff need to promote what justifies it. That’s what gives cloud providers and software vendors the edge over network operators and network vendors, and the latter group needs to think that point over carefully.