Intel reported its quarter, and the numbers were better than many had feared, though sales in the PC market grew only 4% while they grew by 15% in the data center. In some ways, the picture is a bit of a poster-child for the cloud, and there were aspects of the earnings call that also relate to other parts of the IT space.
Data center growth at the high end of the chip line is, I think, a pretty conclusive indication that companies are buying systems structured for resource pooling strategies like virtualization and the cloud. The goal of these pools is to exploit the additional computing power available to manage total cost of operations by reducing the number of systems needed to run a given application set. It’s a part of the cloud story, but I don’t think it’s fair to say that the cloud is driving IT in any sense. In the currently popular cloud model, the notion that the cloud is cheaper than traditional IT, it would have a net negative impact on system sales because you can’t spend more to replace something than that something cost and still say you’re saving cost.
Intel also reported strength in the ultrabook lines, and I think that’s interesting because it may validate one of the key thesis of Microsoft’s initiatives; workers aren’t ordinary consumers. Ultrabooks are tablet-like in form factor but are real PCs, meaning they run everything a PC does. If the world were moving to a completely cloud-dominated future, we’d expect to see less need for powerful appliances; power would shift to the cloud. Microsoft’s Office strategy, which clearly suggests they think personal productivity apps will run on the devices in the users’ hands, plays to that same position. Maybe the cloud is the future, says Microsoft and Intel, but it’s not the future yet. If the cost-savings driver is the cloud’s only stimulus, frankly, I don’t think we’d ever get there.
It isn’t, and another development may prove that. VMware’s decision (or more likely EMC’s decision) to change out CEOs may be an indication that the company is going to step beyond virtualization or building bottom-up resource pools from servers and into a broader cloud vision. The rumor is that VMware wants to get into network virtualization, which means SDN. I think that would be significant, but more as a symptom of a maturing vision of SDN than as an opportunity for VMware to suddenly dominate the space.
SDN has two “home ranges”, one as a foundation for resource networks and another as a foundation for access/user networks. In the former context the SDN has to be paired with cloud provisioning and integration systems, the “DevOps” stuff, and in the latter it has to be paired with security and identity and access rights management. There is no question that VMware could take and likely sustain a compelling position with DevOps, and that they could use that position to generate a conception of an SDN as a cloud partner inside the network. The challenge is that clouds consume services and not forwarding decisions. There still has to be a context for all those forwarding-table changes, a context that adds up to connection services, and those services then have to be presented to the cloud in a useful way.
Cisco may have something in mind here because they announced the purchase of Virtuata, a firm specializing in cloud security. It’s tempting to ask whether Cisco intends to integrate security into a high-level cloud vision, one of its “Architectures” or Cisco’s ONE. The problem is that in the past Cisco has typically bought companies to manage sales relationships; a very tactical mission. Other network vendors could be doing something here, too. I talked with Juniper yesterday about their cloud/SDN plans and at the high level they seem complete and potentially compelling. The devil will be in the details, just as with Cisco. Having all the pieces of a puzzle doesn’t build that picture of the Grand Canyon.
In the midst of this cloud/SDN symbiosis issue there’s the pressing problem of operator revenue and profit. A study released by Telco 2.0 says that data revenue gains are no longer sufficient to offset voice revenue losses and that operators will face major losses by the middle of the decade. I’m not completely on board with the specific numbers, but the theme is consistent with everyone’s forecast (including my own) that ARPU growth in mobile will plateau and begin to fall some time in 2013 and wireline is already going that way. That ultimately hits operators ROI, which means less “I” unless you get “R” back on track. The cloud/SDN symbiosis could link the service architecture the operators are buying into (the cloud) with network infrastructure investment (SDN). That’s why we think this union is the biggest issue for the operator-focused equipment vendors; you can’t be a success if your buyer is a failure.