NFV (Network functions virtualization)

A Big Week for NFV

More than ever before, the industry's eyes are focused on the network functions virtualization (NFV) Industry Specification Group (ISG) that is helping to shake up the telecoms technology sector.

The group, formed about a year ago under the auspices of the European Telecommunications Standards Institute (ETSI) , has battled through its formative months and is now entering a period when it can truly shape the R&D and strategic roadmaps of network operators and the vendor community for years to come. (See NFV Group Finds Its Feet.)

Now the members, of which there are about 130 (mostly vendors), are gathering for its fourth meeting, being held this coming week (October 30 – November 1) in Sunnyvale, Calif., with Juniper Networks Inc. (NYSE: JNPR) the host. The group has little more than a year left to pin down guidance and specifications for the industry (it was always set to have a two-year shelf-life) so it can't afford to waste any time, though that's a tough task for any organization that bases its decisions on consensus.

It's especially tough when there is so much to lose and so much to gain for many of the member companies. While some in the industry are fully embracing the open source model, others are not convinced (and that includes operators as well as vendors). (See One Problem With NFV.)

But there has been incredible progress made in recent months, with the NFV Showcase that took center stage at the recent SDN & OpenFlow World Congress a case in point. (See this whitepaper for the details.)

What's key at the upcoming meeting is that the lead carrier members of the group keep in mind the key questions that they were asking from their initial gathering in Germany a year ago: Can NFV deliver true operational and financial benefits to network operators and will it make life easier, or more difficult, in the long run?

Those are very tough questions to answer, especially the second one. The operators, more than anything do not want to put their combined weight behind a new technology direction that takes them down an operational road to hell that's even worse than the silos and technology lock-ins that make life tough now.

Every hour of these NFV ISG meetings can make a difference to the future of all involved: The group's leaders need to make sure that the path towards a greater good is not blocked by those with less egalitarian ideals.

— Ray Le Maistre, Editor-in-Chief, Light Reading

Page 1 / 2   >   >>
[email protected] 10/31/2013 | 1:51:41 PM
Re: The PoCs are the Key Yes, thanks Tom for taking the time to answer such questions on these boards - much appreciated.

Also, do note that these are the sorts of questions and issues that will be addressed in the first Light Reading University course in November -- see



sam masud 10/31/2013 | 1:02:06 PM
Re: The PoCs are the Key Thanks, Tom. Appreciate you taking the time to reply.
TomNolle 10/30/2013 | 6:01:07 PM
Re: The PoCs are the Key Not dumb at all.  What I've found so far is that the availability of software for use as a virtual function will depend on both the scope of software sources and the restrictions that the NFV implementation might impose.  If you presume that an NFV implementation has special APIs for functions to use to help their deployment and management, then software would have to be customized to those APIs, which means nothing off the shelf is likely to work.  That, in turn, would mean that operators or vendors would have to prepare the software to run.  Likely the first stuff would be things like IMS that are widely desired, and service-chain functions like firewall and load balancing.

We, in CloudNFV, presumed that the goal was to make ALL cloud software suitable for running as a virtual function.  If that's done, then in theory anything can be made to work, but again there are things that NFV applications should have (horizontal scaling, resiliency) that not all cloud software has.  We picked Metaswitch's Project Clearwater as our demo for CloudNFV because it was the ONLY package available that fully realized NFV potential.  However, for basic service chaining most Linux network tools work fine in single-tenant mode for business users, so all of that can be grabbed and used immediately--if you design for it!
sam masud 10/30/2013 | 2:49:42 PM
Re: The PoCs are the Key Hi Tom,

Since the NFV project covers a number of functions (SBC, NAT, etc.), I'm curious whether some functions might get virtualized quicker than the others. If so, then which ones......or are those dumb questions???

Also, have any concerns been raised regarding performance and scalability of virtualized functions?

TomNolle 10/28/2013 | 3:15:44 PM
Re: The PoCs are the Key I think that's a good point right back at me.  I do believe that the operators intend the ISG to be very agile, and it's my contention that the place where this agility has to manifest itself is with the PoCs.  If they get a good PoC process going it can not only lead the work in the right direction, it can connect the work with the broader set of issues that will ultimately have to be addressed by NFV-based technology.  A little nubbin of NFV in a vast device world has to pay back in some way and demonstrate the longer-term and broader-scope value proposition, or you'll never get beyond that nubbin because nobody will see the busienss case.
Gabriel Brown 10/28/2013 | 2:15:22 PM
Re: Let's hear it for the VNFs The EANTC demos shown in Germany the other week were very interesting, agreed. Basically, a public demnstration of what operators are discovering in trials.

Also, Mano!
Gabriel Brown 10/28/2013 | 2:11:18 PM
Re: The PoCs are the Key Hi Tom -- good points. I agree on the importance of PoCs. But I'm not sure that success or otherwise can be hung on the ETSI process. My feeling is that NFV will be a little more anarchic than a traditional specification effort led by a formal standards body.

The ETSI initative itself seems to be designed this way -- supposedly the group will be phased out in little over a year -- and has purposefully left room for flexibilty for operators and vendors to experiment with NFV and adapt their strategies as the target itself evolves. This fits very much with an important objective of NFV: to be more agile.

To be agile, operators need to start agile and stay that way.

But I may well be wrong. As the NFV ISG solidifies an organisational structure and vested interests it will fight to survive and impose its will.
Dredgie 10/28/2013 | 1:54:14 PM
Let's hear it for the VNFs Nice piece. Glad you are calling-out VNF's in this article (like those tested by EANTC and documented in their white paper, here). With good reason, there is a lot of talk (and progress) with regards to the infrastructure of NFV - MANO and all - but maybe that just because people like saying MANO! But anyway, let's not forget the virtualized network functions, without which all of this is for nothing. As Tom is too modest to do it, I'll plug his latest Blog entry: 'VNFs equal What Plus What?' Another great introduction to NFV and specifically, in this case, what it takes to deliver a network function in the cloud. http://blog.cimicorp.com/?p=1513

And of course there's the CloudNFV initiative that Tom is driving for NFV PoC. http://cloudnfv.com/

OK – enough with the link spam! :-)
TomNolle 10/28/2013 | 9:47:09 AM
The PoCs are the Key IMHO the key to the success of the NFV ISG will be the pace at which the proof-of-concept applications are submitted and approved.  Remember that getting a full specification isn't necessary to do PoCs, and also that a full specification isn't necessarily an implementation.  The market could wait until January 2015 for a complete spec but not two years beyond that to get a compliant implementation.  The PoC process was just approved (about 10 days ago) and now it's time to see how fast it can progress.  The IETF has proved that you make the best progress the fastest if you can harness live implementations to guide specification.
DOShea 10/27/2013 | 9:10:25 PM
Re: Working to a deadline Two years sounds about right as far as the state of trials and what at least some of our readers seem to be expecting (as indiciated from recent live chats), but I agree deadlines push a group either way if they don't stay focused.
Page 1 / 2   >   >>
Sign In