Well, Happy New Year everyone. There are obviously positive and negative omens in play for the world, economic and otherwise. Still, I’m a network/cloud strategist and not a doctor or economist. I’m not going to attempt to address broader issues and events, other than to point out what a negative versus positive shift could mean for networking. I blogged at the end of 2021 on the three technology factors that were likely to influence networking in 2022, and now I want to bring all the points together, not by reprising the early blogs but by summing the impacts and presenting an overall view of what to expect.
Network services are at the heart of everything. The Internet is the new dialtone of the world, and the mechanism whereby we shop, sell, support customers and link with partners, entertain ourselves and even learn. The Internet is functionally an over-the-top community, but network services are what make the Internet possible. Those same services are the foundation of business networks, calling and texting, and everything that connects us. The profits of services fund service provider capex, and the nature of the services determine the network plans of enterprises and the personal behavior of consumers. For all these reasons, it’s a good idea to start our discussion with network services.
The problem we have with network services is that return on the investment needed to sustain, much less expand and improve, them is in a squeeze. Virtually every business wants to use network services, and certainly most consumers, but nobody wants to pay for them, or at least pay in proportion to usage. The Internet is the first network service the world has ever known that didn’t settle among participating network operators. While consumers have driven traffic growth, and their unwillingness to pay for usage has cemented in the neutrality rules, as the cost of Internet capacity drops and bandwidth and QoS improve, the Internet looks increasingly attractive as an alternative to traditional business networking, like VPNs. We only have to look at the growth in SD-WAN to see that.
The fact that network operator profit per bit on basic connectivity has shrunk dangerously close to the point where ROI isn’t satisfactory is clear. Many operators have started to move out of their traditional service markets to other geographies or other service areas. The good news is that we’re not going to see network operators exiting network connectivity services. The bad news is that we are likely to see a curtailment of investment there, but fortunately not in 2022.
The original value proposition for networking, the value proposition that drove everything up to the point where public Internet got popular, was connectivity. A connectivity service is valuable to the extent that it can connect a lot of stuff, and that tends to value a broad network. Consumer Internet and broadband changed things because the goal was to connect to OTT resources, experiences, that became increasingly cached and local. The natural evolution of content services results in a metro-cache focus, even if we ignore other trends that drive metro change. Operators don’t seem to see this, and the likely reason is that they don’t want to see a future dominated by metro-centric services. They’re connection people and they want to stay that way.
What’s keeping operators committed to connectivity services isn’t public-spirited thinking, it’s more like tradition or inertia. Operators have over a hundred years of both invested, and they’re no more likely to jump up and become over-the-top players than a farmer would be likely to decide to become a miner. In fact, what we’re seeing now is that operators are doing what farmers would typically do if somebody discovered mineral wealth on their land—they outsource the new opportunity. 5G has created the closest thing operators have seen in decades to a “new service opportunity”, but they don’t see the real opportunity, which is edge computing or “carrier cloud”, as one they can run with. So, they’re doing deals with the cloud providers for the edge and hoping for both “thing-connect” revenue and some sort of QoS windfall on connection services.
The cloud providers aren’t all that interested in the kinds of “new service opportunities” that operators think 5G could create. They understand that “thing-connect” charges would do nothing other than invalidate the whole IoT service space, and that nobody wants to pay for QoS when demonstrably the Internet and much of cloud computing don’t require it. Cloud providers do see reality here, and want to seize the higher (in revenue and profit margin terms) ground. They are interested in the 5G hosting opportunities operators are presenting, in edge computing partnerships, and other relationships that involve operators paying for cloud hosting. They don’t want operators to build carrier cloud, any more than operators want it.
What this likely means is that operators’ investment in carrier cloud in 2022 will be negligible, and that “new services” offered in 2022 will likely be simply face-lifts on connection services. That includes 5G, and Verizon’s decision to defer 5G Standalone, meaning 5G Core and the slicing features, until “2022” likely means that they don’t really know when or if they’ll push full 5G. Whatever we see in new network services, including edge computing and IoT, is more likely to come from cloud providers than from operators.
If operators want connection services, though, they do have the avenue of SD-WAN and SASE to think about. SD-WAN is a virtual-network overlay on IP and the Internet, targeted initially at sites where MPLS VPNs were either too expensive or unavailable. When the Internet, at least at the consumer level, was a dial-up phenomena, a separate business network framework made sense. Today, with the Internet charged with video delivery and delivering much better QoS, that’s probably not the case. I think operators in 2022, fleeing OTT-level “new services” like edge computing, will push an integration of SD-WAN, NaaS, and SASE. They’ll leave the rest to the cloud providers.
If the operators aren’t likely to be offering services that compete with cloud providers’ services, the opposite is not true. By the end of 2022, I think all the public cloud providers will be offering things that will effectively compete with operator services. The starting point will be intra-cloud networking designed to encourage enterprises to create a global front-end in a single cloud, linked to multiple data centers within the cloud itself.
What’s interesting about the intra-cloud network is that it will have a lot in common with the network needed to make metaverse a reality. Building a metaverse demands edge computing in metro centers, and meshing of metro centers with low-latency pathways. Such a network is fiber-intensive because rather than feeding traffic to a core network, each metro center directly feeds fiber to other centers. The core of the network, if that’s an appropriate term, is much more optical than traditional IP.
Even though “carrier cloud” isn’t likely to drive edge hosting in 2022, and it’s not likely that the metaverse will by itself stimulate the metro-mesh future that I think will sweep networking, we will likely see more pedestrian forces of competition among cloud providers lay the groundwork for radical changes. The cloud providers see what operators can’t or won’t see, which is that connectivity is just a service on-ramp whose only useful attribute is cheapness. There’s no money in offering that sort of thing.
What “metaverse” as a concept does is focus the most aggressive competitors on the metro, and on the fact that service value ends up in the metro, however you define “service” and whoever the “service provider” really is. Metaverse also focuses on the fact that what makes metro different in the future is the integration with edge computing. If the network operators do in fact cede the hosting to the cloud providers, they’ll self-disintermediate on the most important future trend. Hosting wins, so cloud providers win.
The industry as a whole may be starting to see the light on this. Juniper posted a podcast that reflects the view that the public cloud providers will end up providing edge hosting, even out to the cell sites. Operators would then have to try to make QoS valuable, but that presupposes that the cloud providers would even consider buying connection services for edge and metaverse from operators. If you can afford to invest in a cloud or edge data center, you can probably absorb the cost of linking it via fiber to some or all of your other centers. Particularly if you can sell “cloud networking” to enterprises to replace their VPNs.
Cloud-provider-dominated edge computing would lead to cloud-provider-dominated metro, and that IMHO would mean that the cloud providers would likely mesh edge enclaves, regional centers, and other hosting points with fiber. That would essentially displace the packet core, moving most routing capacity to the metro areas, and making metro the focus of electrical networking capacity overall. It would also create resources whereby the cloud providers could expand intra-cloud networking that they’re already starting to offer, and offering a variety of cloud-and-edge-related VPN services.
Now, in closing, let’s look at the question of health and economics. It seems clear that Omicron doesn’t cause as many serious cases as Delta, and it’s also clear that vaccination has reduced the risk of serious problems even further. We’re coming to terms with a form of COVID that is more like the flu, and that means we’re accepting a level of infection risk as the price for minimal social and economic disruption. It’s those socioeconomic factors that created lockdowns and business issues, and so it appears that 2022 doesn’t have another wave of impacts in store for us, just another wave of virus. Barring another variant with greater impact, we should see 2022 as a year we shift more toward “normal” behavior. The forces that shape networking in 2022, then, are already visible and operating on the markets.
Can the cloud providers do networking, leaving operators with little or nothing beyond the access networks, which are the lowest-margin and therefore least-competitive pieces? If they do, then it may mean that some form of public/private partnership will be necessary to maintain investment in broadband access. The form may vary, with some countries perhaps opting for something like Australia’s NBN and others creating broader subsidies, but the end result would be the same. Value may be tapped off the pure OTT players, but it will come to rest in the public cloud and not carrier cloud. This year, 2022, is likely the last chance for operators to empower themselves at the service level. The most important question of the year is whether they’ll see that, and do it.