It’s “recap Friday” again, and there are a lot of little items that (as usual) add up to some significant industry trends. Facing important truths about something you depend on is never easy, and it looks like the industry as a whole and some companies in particular are now going to have to do that.
Networking equipment numbers have generally been soft in this earnings season (Cisco doesn’t report till next week), and I think this frames the central issue of the industry—you can’t expect buyers to spend more to generate less ROI. Vendors have historically believed that connectivity and traffic-handling were divine mandates that enterprises and operators simply had to meet at any cost. It was up to the buyer to make the business case. Well, we can see where that’s been getting us.
Not surprisingly, the need to improve ROI has driven both enterprises and network operators to look at cost management as a partial solution. This has driven some vendors to take a completely cost-side view of the future of the industry—Juniper is probably the best example with their TCO stories. The problem with that is that the least costly thing to do is to not buy anything. I’m looking at an old D-Link hub I bought a decade ago for probably twenty bucks, and it’s still churning out bits. Are we saying that consumer network equipment is more reliable, has a longer useful life, than carrier gear?
Or, like most vendors, you can add some enhanced features to your products and tout them as a justification for churning the network. This idea also fails the school of history. Novell, once the darling of PC networking, dumped because they didn’t recognize that at some point buyers didn’t need more bells and whistles added to file and printer sharing. Same here in networking.
The vendors who need to be the most concerned here? Likely the giant equipment vendors who are tied to a portfolio of low-margin equipment kept afloat by the single bright light of mobile services. 4G LTE competition among operators has given some vendors a lease on life, but we just had what I think was the first truly insightful IMS open-source announcement from Metaswitch. How long will it take before you can buy IMS/EPC components in open-source form and then (via the operators’ NFV initiative) deploy them on commodity servers?
And why are commodity servers important? Because the quarterly numbers are demonstrating that proprietary servers are just not selling. We are seeing in servers, as in networking, the fact that one box is pretty much like another—they’re samples on the paint card of life. Thus, price competition is all there is there. IBM, Oracle, HP…everyone is seeing power go up and price per unit of power fall.
Why this is happening goes back to that ROI picture. Software is the value of technology, not the boxes. Software has low inertia, can be replicated endlessly at low incremental cost, and is the component that matches technology products to human behavior. Hardware runs it, connects it, but doesn’t make it valuable—it’s the other way around. The server, and the network, is a software platform. That’s what SDN and NFV are really about; bringing both servers and networks into a functional union to take the next step in supporting what will have to be a software-driven revolution.
The problem is that this sort of revolution is hard for people. I happen to have been a software architect in my life prior to consulting, and so I’m comfortable with software. I’ve talked just this year with hundreds of good technologists on the network side who have no intuitive understanding whatsoever of software. I’ve talked to software executives who have absolutely no understanding of networks. We have market forces driving us to a unity of concept with precious few people to do the unifying. That’s true with vendors, with network operators, with standards-writers.
When Cisco announces its numbers and makes its comments on the quarter next week, what I’m going to be looking for is an indication that they have a plan for the software side of the networking revolution. They have taken steps like their ONE stuff, but that’s not all that different from the API position Alcatel-Lucent took, and may be un-taking as we speak. What is necessary isn’t to expose network assets, it’s to leverage them and to use software to build a bridge to utility. We want more network spending? Then create more network value. It’s not Cisco’s “Internet of things” or Juniper’s TCO. The former says that the solution to voice revenue was more mouths, not what those mouths can say that’s useful. The latter says that the solution is to acknowledge that there will never be a new network application, so we’ll sit on the shrinking island until global warming drowns us. So watch Cisco this quarter and through this summer. They have the best assets for the new fusion of networking and IT, and if they can’t make a go of it, things are going to go hard for the industry in 2014.