The Carrier Ethernet Forum isn’t where you’d expect to find things like SDN and Network Functions Virtualization, but there was plenty of both there, and that’s likely an important step for both the group and the two technology initiatives. SDN and NFV are getting the buzz, enough that groups like CEF think it’s important they jump on the bandwagon. That could be the start of the classic bandwagon effect for SDN and NVF.
The challenge, as an AT&T spokesperson said in her talk, is getting SDN defined in a way that actually meets its goals. As I’ve pointed out repeatedly, OpenFlow is a tentative first kiss in the SDN marriage, and what’s getting married (whether we realize it or not) is the cloud and the network. I think a lot of people at CEF know that, which is why the body is looking harder at cloud-ready APIs. But CEF illustrates the risk of Balkanization of the SDN standards by the introduction of a host of standards groups.
Despite the risk, the CEF interest in SDN might be critical because the reality is that SDN is likely to be very much an “Ethernet” technology. First, virtualization principles overall are more likely to start in the data center because of the needs of cloud multi-tenancy. Second, if you’re going to dumb down a network device that forwards packets and slave forwarding to a central process, you’re left with something that’s little more than Ethernet interfaces connected with a simple forwarding fabric. Third, the applications of SDN in the metro will clearly be Ethernet-based, and finally Ethernet and optics will be creating the evolved core network model. I’m not saying that IP is dead; rather, it will be the “service protocol” but not so much the transport protocol. You can see these trends already.
The overall economic questions may create additional pressure here. There’s no doubt that both network operators and enterprises are parceling out their capital dollars with great care, due in no small part to the fact that the US fiscal cliff and the Euromess remain open questions. For tech vendors, the fact is that the best thing that could happen to them would be for there to be a sudden attack of bipartisanism in Congress. If US risk factors were off the table, I think we’d see a major surge in the US economy that would likely restore the US share of the tech market for both enterprises and network operators. That would have the effect of pushing sales and opportunity ahead of any possible massive changes in network technology, and it would therefore delay the adoption of SDN/NFV principles. On the other hand, a protracted problem with US and European capital budgets (likely, sad to say) would create enormous pressure to develop cheaper network infrastructure, which could advance both NFV and SDN quickly. Once they advance, they’ll never recede.
You can see that vendors are under pressure for sure. Cisco is pushing the notion of WiFi as the real mobile future because it has no 4G position to rival Alcatel-Lucent, Ericsson, and NSN, and nobody without mobile has much of a shot these days. Alcatel-Lucent and Ericsson are both cutting staff to cut costs. Juniper has dumped one of its big suppliers, Plexus, and the story is that they’re also trying to cut costs. All of this reflects a market that has lost pricing power, and that’s a market that simply has to consolidate. But SDN/NFV may show the way for any vendor brave enough to step up RIGHT NOW. You can follow the other lemmings over the cliff, dear vendors, or you can hunker down a safe distance back from the edge and build a veritable market castle in a field that will be free from competitors. To me, the choice seems obvious, but any corporation is filled with people expert at telling management what they want to hear.