SDN architectures

Setting the SDN Agenda

DÜSSELDORF -- SDN & OpenFlow World Congress -- When I used to think of Germany, I would think "lovely people, amazing cake." Now, I also think of SDN, because this event has made its home here for the past few years and doesn't look like it's leaving -- at least not while Deutsche Telekom still has SDN and NFV at the heart of its next-generation network strategy (which it does). (See Nothing Is Sacred, DT's Clauberg Tells BTE .)

So what are going to be the hot topics of debate this week? I've been here a day, sitting in on the Open Networking Foundation (ONF) workshop and chatting to a number of companies with a vested interest in SDN's future success, and there are a number of debates likely to rage all week:

1. Turning open source code into carrier-grade, supportable technology: Everyone can see the benefits of the open source movement and the role it can play in the SDN sphere, but who is going to turn the developments of projects such as OpenDaylight into "hardened products" (for want of a better term) that can be tested, supported and deployed with confidence in wide area networks? Let's watch the vendors form a disorderly queue this week.

2. Which standards and specifications are worth the time of day?: Related to item 1 in many ways, the question of which specifications/standards really matter to network operators when it comes to SDN and NFV is heating up as more and more industry groups announce they're developing code for all manner of virtual/programmable deployment scenarios. It's clear that the specs and proofs of concept (PoCs) developed by the European Telecommunications Standards Institute (ETSI) NFV Industry Specifications Group (ISG) are important, as is the ONF's OpenFlow protocol for SDN, along with work being done at the Optical Internetworking Forum (OIF) , OpenDaylight and OpenStack . The Internet Engineering Task Force (IETF) is a given (in my book). But others? Is anyone other than Alcatel-Lucent (NYSE: ALU) backing TOSCA (Topology and Orchestration Specification for Cloud Applications)? Let me know on the message boards below which others should be added to this list, and which ones are has-beens.

3. Multi-layer provisioning using SDN-enabled controllers: We've reached the stage where being able to configure and provision optical and packet network layers using a single controller is not only a carrier wish-list item, but a reality. But which vendors will have this to market first and with the right set of functions and tools to tick all the operators' boxes? Best buddies Ciena Corp. (NYSE: CIEN) and Ericsson AB (Nasdaq: ERIC) reckon they have this licked (and have a "live" demo to show off) and so does Coriant . By the end of the week that list will be longer -- this is a real battleground.

4. The cultural shift: Light Reading has been banging on for years about how the telcos need new processes and skills to deal with the integration of IT and telecom, of which SDN and NFV are now the prime and most high-profile manifestations. But do any operators have the skills and organizational nous to make the most of SDN's and NFV's potential? The jury isn't even out, in my mind -- currently most of them do not. This will be an ongoing debate this week, even if I'm the one who has to fuel it.

5. The business case: Gone are the days of deploying a new technology because it's cool, hot or sick (or whichever word means "really good" in your 'hood). Now it needs to be driving down costs and creating new revenue opportunities. So what are those in the SDN and NFV worlds? That's a debate that simply has to happen this week, and every week.

Keep an eye on our reports from Germany this week to see how these play out.

— Ray Le Maistre, Circle me on Google+ Follow me on TwitterVisit my LinkedIn profile, Editor-in-Chief, Light Reading

[email protected] 10/15/2014 | 12:13:49 AM
Re: Thinking Beyond "Consensus" Thanks Tom - good points all.

Seeking industry comment on the merits of the various specs/standards/models is certainly something we aim to do in general but certainly something we can do in a more dedicated, specific feature. Certainly one to follow up this week.
TomNolle 10/14/2014 | 3:46:12 PM
Thinking Beyond "Consensus" I think, with respect to your point on which standards are important, that the question has to include "consensus" but at the same time go beyond it.  I think most people agree that both SDN and NFV are in their infancy, which means that the number of people who are exposed to the deep details are very limited.  This limits the value of consensus appeal for standards, IMHO.  TOSCA is a good example.  Do we want standardized descriptions of deployable network services?  If "Yes" then it's hard to see what options other than TOSCA exist.  Perhaps LR would be willing to look at the various standards/sources and solicit comment on why they are important?
Sign In