How Mature is your Network?

Date: Wed, 11/12/2014 - 10:11

How Mature is your Network?
MEF SDN & OpenFlow World Congress roundtable, Dusseldorf

Chaired by: Michael Howard, Co-Founder and Principle Analyst Carrier Networks, Infonetics Research

Panellists:   
Phil Tilley             Managing Director, Cloud Strategy and Solutions, Alcatel-Lucent
Prayson Pate        CTO & Vice President of R&D, Overture Networks
James Armstrong  Executive Vice President Network and Applications and Infrastructure, Spirent Communications
Ulrich Kohn           Director of Technical Marketing, ADVA Optical Networking
Marc Cohn            Senior Director of Marketing, Ciena Corporation
Andrew McFadzen  President, OVCC
Gint Atkinson        Vice President, Network and Strategy & Architecture, KVH
Chris Purdy           CTO, CENX
Neela Jacques       Executive Director, OpenDaylight Project

Michael Howard
So this is the voice of Michael Howard - Co-Founder and Principle Analyst at Infonetics Research, 25 years in the business here.  I know almost everybody in the room.  Most of you I've known for quite a few years.  So with that, I've been following just SDN and NFV for the last three and a half or four years, starting with a stealth introduction to some little company called Nicira.  It's no longer independent and no longer stealth.  But anyway, that was the start of it. 
So with that, I'd just like to turn it around to head round the room.  Phil.

Phil Tilley
Okay, yep.  Phil Tilley, I'm with Alcatel-Lucent, leading Cloud marketing.  So really in Alcatel-Lucent, there's two parts of this, it's managing the integration between - or the positioning of the Nuage Networks on the SDN side and CloudBand on the NFVi and then all the virtual functions we also have as well.  So an interesting role for managing all the elements together.

Prayson Pate
I'm Prayson Pate.  I'm the CTO and SVP of engineering at Overture Networks.  We are coming at this from a little different angle than some of the others.  We've had a long history in the carrier Ethernet business starting about two years ago we changed our focus from building boxes to really enabling our customers to deliver services.  We did that by embracing these new technologies of SDN virtualisation and NFV and intersection.  And it's really interesting times going on and I'm looking forward to the discussion.

Michael Howard
Hopefully there's something edgy about your products. James

James Armstrong
I'm James Armstrong.  I'm the Executive Vice President of Network and Applications and Infrastructure at Spirent Communications.  I've been with Spirent for six months.  So relatively new there.  But what is certainly high on our priority list at Spirent is the testing and bringing assurance to these virtual networks.  We talk about elasticity, we talk about rapid spin-up of new devices and services and the flexibility that comes with it.  But how do we get the assurance that it's going to work properly?  So there's a lot of investment from our customer base in that area.  We're working hard at continuing to provide visibility in the - and understanding the parameters around giving that kind of assurance.

Ulrich Kohn
My name is Ulrich Kohn.  I am with technical marketing at ADVA Optical Networks.  So for a year and six months only, longer than you in your role.  ADVA's main focus right now is putting programmability and openness into carrier Ethernet and optical connectivity solutions.  So we have two main areas of engagement, [unclear] and on top of that [unclear] and flexible optical [network] solutions.

Michael Howard
Okay, so - Marc.

Marc Cohn
Oh, yeah.

Michael Howard
The press aren't going to introduce themselves.

Marc Cohn
I'm Marc Cohn, from Ciena Corporation.  I'm Senior Director of Marketing and I work in the SDN and NFV software area within Ciena, leveraging our industry leadership position in tackling optical infrastructure.  I also have some external affiliations.  I chair the market area for the Open Networking Foundation, so I've been involved in [unclear] since the beginning of the ONF.  I also sit on the board of directors of OpenDaylight, the open source framework for SDN controllers.

Michael Howard
So it sounds like you don't need the work, then, here.

Marc Cohn
No, I don't need to sleep, Michael.

Michael Howard
Andrew.

Andrew McFadzen
Yeah, hi.  I'm Andrew McFadzen from Orange, Orange Business Services.  So I'm in charge of global marketing for network solutions.  I'm President of the OVCC, which is the Open Visual Communications Consortium which is a group of about 25 vendors and services providers providing B2B video services, but I'm also the Chairman of the MEF.

Michael Howard
Gint.

Gint Atkinson
I'm Gint.
With KVH from Japan, a sister company of COLT in Asia.  I've been in the business now just about 30 years.  I came to KVH from Ciena, World Wide Packets and before that T-Mobile.  What I did at KVH when I first came a few years ago was - we were the first 100 gig service provider in Japan and were the market leader for 100 gig services in Japan.  We also were doing SDN four years ago, but inside the datacentre.  We decided that the way to go with our new service architecture was to be very flexible and dynamic and elastic, just like we were inside the datacentre, and take that into our connectivity services and play into the cloud connectivity space.
So today, along with COLT, we're one of the first providers to offer SDN-based WAN connectivity, going all the way into cloud providers, for example, Amazon, Direct Connect.  So our long-term vision is to be able to orchestrate service provisioning all the way from Amazon, all the way across the cloud, all the way into SoftLayer and Azure and do service mashups, because we're never going to be able to offer compute storage at those price points.  So why don't we just orchestrate the whole thing?
This will allow a lot of our customers that are in dozens of datacentres to just be in a few key datacentres and get everything they want from their trading partners all over [our] datacentre and our connect network.  So you can check out our DCNet service, it's pretty fun.

Michael Howard
My understanding of your position is sitting in the hot seat, given your boss?

Gint Atkinson
We have a very exciting CEO, and Ted is extremely aggressive, adventurous, visionary and also technical.  So he can get into every detail of strategy from the customer side all the way through technical and operational.  So we're able to move fast.  I just got an email a week ago from him and he said, develop and launch this new service in eight weeks.  So DCNet, our DCI service offering, was conceived and launched in under six months.  That includes building out over 100 PoPs across Asia and terabits of capacity all over the place. 

Michael Howard
So now that you've done that in six months, the next project is eight weeks.

Gint Atkinson
Exactly.

Michael Howard
Okay, I understand.  Chris.

Chris Purdy
Yeah, I'm Chris Purdy.  I'm Chief Technology Office for CENX.  We are a software company that sells service orchestrators.  So just as Gint was talking about the need to orchestrate, we would be trying to dupply service orchestrators to help the orchestration.  So by service orchestration, think of services as being connections of all types going from anywhere to anywhere else.  Those connections are becoming much more dynamic in terms of the [apex] connect right into the datacentre providers [then] connect with cloud providers so it's much more complex.  But they also much go over both the existing infrastructure and the new SDN, NFV infrastructure.  Because in the real world, nobody's going to just put in entirely new stuff, they're going to - for a very - there's going to be a very long transition associated with that. 
So our - and also by service orchestration we talk about the entire life cycle.  So it's not just about establishing the connectivity, it's understanding the operational state of that connectivity, is it meeting its SLA and then ultimately it's complete life cycle until its eventual disconnect.  So that's our mission.

Michael Howard
Isn't the CTO or CENX, Carrier Ethernet? 

Chris Purdy
Yes.

Michael Howard
So you focus on carrier Ethernet services?

Chris Purdy
I'd say that our first - so originally, we were focused 100 per cent on carrier Ethernet.  We actually started in a very different form.  We were originally going to be a service provider, actually doing Ethernet exchanges.  The notion was it was very complex to buy and sell carrier Ethernet services between parties.  So the challenge that we faced was that we got a lot of people that physically would - the buyers and sellers would physically interconnect to us.  But most of our investment was in the software of how you ordered between parties, how do you actually do the provisioning between parties, how do you sectionalise between parties, how do you actually measure whether or not the other parties are meeting their SLAs?
What happened was, when we were going to do a demo of our software to the service providers, they would say, well you can't get your exchanges in enough  of the places where you really need to be but I'm interested in your software.  So long story short, we actually ended up making the pivot to focus on that.  So I'd say initially, fully carrier Ethernet, focus on carrier Ethernet and now pure software carrier Ethernet.  But we're now going into pretty much - up into the IPVPN and [unclear].  So I'd say our current focus is really orchestration of connectivity services, advanced data service but we say that really our first is connectivity in general.

Michael Howard
Neela.

Neela Jacques
Sure.  So my name is Neela Jacques, I'm the Executive Director of the OpenDaylight Project.  For those of you not familiar with OpenDaylight, OpenDaylight is the leading open source project in the networking industry.  To describe a little bit about why OpenDaylight exists, you kind of have to back up a couple of years ago and realise that there's something very different happening in networking that has happened anywhere else, which is that as we look at SDN, the fundamental transformation of networking, we see a situation where almost the entire market has jumped on and said, yes, we need this. 
Unlike things like virtualisation when in 2005 the vast majority of people said, we don't need virtualisation, if we do, it's going to be in file and print servers.  With SDN, everybody's like, yep, I need a programmable network, this is critical from a carrier, this is critical to NFV.  So from a [hype cycle] standpoint, SDN's up here.  From an adoption, it's down here.  Less than five per cent of people have actually deployed SDN even though 95 per cent say that they want to deploy within two years.
This actually, unfortunately, hasn't changed that much in the last couple of years.  It's changed a little bit.  So that immediately begs the question why?  If we have a need and we have low adoption, is it that we don't have solutions?  That's the natural one.  Maybe nobody's making an SDN solution out there.  Unfortunately, that's not true either.  Because every single vendor in this space has an SDN solution.  At the heart of every SDN solution there's a controller and everybody has their controller.  They all roughly do the same thing.  Unfortunately, they do it in slightly different enough ways that none of them are truly interoperable. 
They're very interoperable when we're talking about PowerPoint, really good at putting nice logos next to each other and putting pretty arrows.  But when you try and deploy them, what you find it it's a system as if every airline in the world had their own air traffic control system and they all - one used the NATO alphabet, the other actually just speaks in Russian.  Some of them use the same words but maybe in different ways.  Think English people talking to Americans.  I apologise if there's any Brits.
So really what we've seen in SDN is that the market is crying out for a platform.  Since I'm throwing out analogies here, one last one, imagine a world from a customer perspective where there is not only VHS and Betamax, but 28 other offerings.  How many - I certainly skipped the laserdisc revolution because I didn't know if it was going to be a standard.  We have that situation where nobody's adopting because what people want isn't fundamentally a controller, what they want is a programmable network.
Outside of a few people who can say, I'm all Cisco, and so it's really easy - if you're all Cisco and you want SDN or a programmable network, you got really excited about [1PK] a few years ago and now you're super excited about [unclear] or ACI.  On the other extreme, you've got people who are really excited about overlays and buy into Martin Casado's vision that you don't have to worry about the underlay.  The network just is and happens, and we're going to do everything up in software.  So those are the two poles.

Gint Atkinson
And that's going to run really fast.

Neela Jacques
And it's going to run super fast and be really easy to manage.

Gint Atkinson
Yeah, and all of my customers are really going to like the SLA that we can support with that architecture.

Neela Jacques
Right.  In the absence of that, we're crying for a platform.  But what is a platform that can take the stuff that we already have, the stuff that's shipping right now and the stuff that's going to be created in three years, we don't know what it is?  So IBM had the forethought about two, three years ago to say, we're probably not going to own the platform and we're not going to get everybody to take DOVE and make that the leading thing, let's see if we can convince - con the rest of the world to work with us in building a common platform for everybody.
So that's what OpenDaylight is, it is a common platform for SDN, to support NFV also.  We have, today, over 280 developers working on the project, we've got over 40 members.  Pretty much every single major firm - if you're in the networking industry, at pretty much every level, certainly through layer 2, layer 3, but increasingly L4 through 7 also, you're a member of OpenDaylight.  We have over 10 major companies announce products based on our project, two major releases out the door.  In fact, this morning, we just had an announcement that Dell has just upgraded to platinum and is dedicating over $10 million worth of engineering resources onto the project.

Michael Howard
Great.  So now we've introduced the speakers and we've been speaking.  So we've got some information out there for the press members.  But we have some questions that we're going to discuss here.  What we heard here in descriptions for the most part were the nuts and bolts of how do you implement NFV or SDN.  But what's the end game or what are we really trying to achieve?  Our service providers on the panel, we've got together beforehand and figured out some questions to ask. 
So the first one we're going to discuss then is a carrier question.  It says, how can SDN and NFV - or - help service providers grow revenue from their existing and the new network services and how can they drive margin improvement?  That's a long way from technology.  So, Andrew, this is your question.  I don't know if you want to start off with answering it?

Andrew McFadzen
No, I'll go first because I'm from - completely non-technical.  So I asked the question so I can start by discussing part of my answer.  I think as a service provider that's been operating and providing a global network for 60, 70 years, we're looking at this very carefully to see what's the opportunity.  Because we've got a lot of kit out there in the market, delivering the service by and large that the customer is asking for today.  Why would we replace that?  When it's not depreciated.
So that's the kind of starting point.  Does it save us money?  I.e. does it improve our efficiency and productivity?  Or is there an opportunity for incremental business as a result of new services that can be generated by either adopting SDN or NFV?  I think it's fair to say that we start off in the camp looking at it as a potential way of saving cost and therefore improving margin.  I'll use an example of NFV.  Today - and I explained this probably in the first session for the press - Orange Business Services' core business is providing wide-area networks for enterprise customers.  So if you take a typical customer, Siemens, they've got 1500 locations on their wide-area network around the world, some of which are very large, some of which are very small.  Each of those locations has a fairly complex piece of kit sitting on it which is a router.
Now, in order to implement that site, we'd have to send an engineer to site, we'd have to ship the box, we'd have to go and repair it when it gets broken.  That truck roll, as the Americans seem to like to call it, that cost of fuel, engineering and delivering that box to site is a pretty significant part of my operational cost.  So if network function virtualisation means that I can take all of the functionality of that box and put it in the cloud, and therefore have a very simple network connection at that customer's site, which potentially can be provided by the guy that I'm buying the local access circuit from, then that saves me a lot of money. 
So from that perspective, as I say, we're looking at NFV, and in that particular instance for that particular application, as a way to save money.  I think as a business opportunity - and again, we were talking about it at coffee - I think it depends what you put into NFV.  To me, it's virtualising network functions, it's not offering access to cloud services, that's something else.  But if I take the example, today, what I would consider as valid as network functions are things like the connectivity so routing, security, optimisation.  They're things that we provide as part of the network service.
If I look at it today, a lot of the customers that we provide the network connectivity for, maybe they go to an integration service provider to buy their firewall or to buy their optimisation device.  Now, if I can provide that as a virtual service then that's an incremental revenue opportunity for me, albeit that it's at the cost of one of my competitors who's an integration service provider.  But as a managed service provider, that's an incremental business opportunity for me. 
So - and in terms of SDN, I think again, we're looking at that as a way to operate the backbone network, the core network more effectively.  We see it - and we were talking a bit earlier - if I look at it from a customer's perspective, what are the main criticisms that they have of a service provider today?  We're too slow, we're not very flexible and we have a very manual way of working with our customers.  So to me, if through the deployment of SDN, we can get to a point where a customer can actually do self-provisioning in real time or near real time then you've ticked two major boxes.  The question is, how much is the customer willing to pay for that privilege.  Is that more than we charge them today as a managed service?
So - and this is where, to me, the jury's still out a bit.  So maybe at that point, I'll...

Michael Howard
Well, you also have competition in your markets too.  Whereas the competitor might come in with a better offer and a more flexible offer and then you lose out.  Then it's not just keeping that one customer.

Andrew McFadzen
No, you're right.  It's about customer retention but also - and again, I think for us particularly - obviously in France we're a very large provider - I was going to say dominant but you might write that down.  We're a big market player in France.  But outside of France, these sort of services and capabilities will give us a competitive - potentially, a competitive advantage to win new business outside of our home market. 

Michael Howard
So as a debate or discussion point, does anyone else - can you offer what a solution might be for them?

Phil Tilley
[I'm not sure about this], I think what's really important there is I think, what you've highlighted Andrew is, there is a difference on what you expect from NFV and what that brings to the table and what SDN brings to the table.  Although they are linked and the technologies are related.  But also I think the other thing that comes in is, associated with NFV, is the virtualisation of the network functions.  So the VNFs, or what we call the VNFs.  Now you talking about the enterprise example, but importantly I think as well is the residential or consumer and the [voice] example. 
So one of the things we're seeing mobile services and in particular VoLTE, voice over LTE which is being launched in the US.  That is definitely all based on virtualised software, or being based on virtualised software.  So historically it combines IMS which would have been tied to the platform and mobile packet core which is again tied to the platform.  The problem is, IMS and LTE are lots of software components that haven't really worked together particularly well.  I think as you start to virtualise them, you actually also introduce the opportunity to start interfacing and bringing the APIs to interconnect those software elements.  So you now start having - re-engineering some of the virtual functions which actually allows APIs to grow - to be developed to now start opening up the ability for third-party integrators to start developing new applications, new services on top, where historically it's been the operator has developed those services, whether it be a voice service or a mobile service or anything else.
Now you've got integrators starting to take voice capability and put that in web pages and because it's virtualised, you can start scanning up and down those quite easily and start onboarding new functions.  So I think in the consumer marker, you're starting to see some new benefits as well; time to market, flexibility, really comes true.  So I think you can't forget the virtualisation of the functions as well as the orchestration of those functions and the networking SDN element to allow you to distribute those functions.

Prayson Pate
So I want to highlight one point that you were mentioning about the new services and we - I heard the term mashup a couple of times.  I think that going back to the beginning of the question, we talked about growing revenue.  To me, that's the most important thing here.  I saw a very interesting article talking about - this was US service providers.  It said between 2008 and 2013 they grew revenue by 13 per cent.  That's pretty good.  At the same time, they were able to cut payroll by two per cent.  So they're already being able to do more with less.  But even with that, their margins are getting worse and worse because the demand is so much more.
What's really interesting is that most of that revenue growth came from the wireless side.  So what that really means is they're driving revenue from the wireless side but they're also driving more demand back onto the wire line.  So they're growing revenue but they're not really growing margin.  So as we talk about these solutions of NFV and SDN, to me what they're really doing is enabling and environment where we can create those new services.  I think that again, what you're talking about with the cloud, I think that's a huge opportunity for providing dynamic services that connect users to new facilities at a lower on-demand cost point.  But you can't do that without effective control of the network.

Gint Atkinson
Yeah, so let me jump on the revenue margin side quickly.  So clearly we - with NFV and SDN and orchestration, we can get really exciting, brand new network functions and deploy them way faster because we're talking about virtualised network service platform.  The network state model has changed a lot more.  Because it's virtual, we don't have the same type of network state representation problems when we're growing our network that we do in a more traditional network.  But we still have so much infrastructure out there.  KVH is getting really, really lean in a way of all the classical BSS, OSS layers that they're pretty tight. 
So a lot of the industry is using Oracle UIM, formerly Metasolv.  I'll give you a good example.  Talking in the same operations people, we have a customer and the customer has two different kinds of SLAs for multi-point connectivity.  Our layer 3 multi-point connectivity is - the highest grade is almost exactly the same as our lowest grade - actually, I don't want to say lowest grade - our middle grade ELAN [unclear] service.  Now, our operations team when we're launching - we take these services and at that level, when you're talking multi-point service, to a customer and you think of their buying criteria, a lot of the customers aren't going to care layer 3, layer 2; give me the most bang for the buck.  You're going to say, well both of these meet your requirements, here is the slight variations.  Do you need to allow transparency well then maybe you don't need layer 2 and maybe you're going to like a lot of layer - what private layer 3 IPVPN's going to give you.
The problem is, every single operation is going to be the same.  When you go to launch that multi-point service, the operations team is going to say, if I get an alarm on this service, I need to know exactly all of the resources allocated with that service.  You scratch your head and you know that our layer 3 services, we've got NNIs with other carriers so we can reach out to parts of Japan and other parts of Asia where we don't have physical plant.  [I'd say] how do those operations guys deal with a broken multi-point service, and they don't know what racket's traversing.  Well, those operations folks, here's the rule, layer 2, we have to know.  Layer 3, we don't care.  Because those layer 3 guys, they're Cisco guys, they're Juniper guys and they come with this culture and heritage.  Then the guys that are doing operations at layer 2, they come from the Optical, SDH world.  
So you have these different cultures and you have two variations of a multi-point connectivity service.  You have very different behaviours locking in.  So how do we get over this?  Well for years we've used OSS and automation to create that transparency.  But in the end, in this example, you can see - you can automate everything but they're still going to bring [up] the beef, where do I push the button to see all of the physical and virtual resources that are allocated.  When we've virtualised everything, it helps, but still, eventually, you may have to drill down to that.  You have - we have a small ops team, only 300 people.  Can you imagine what it's like with 3000 or 30,000?  When I was at T-Mobile, really big staff.
So I think we're back to some classical problems.  Business process, cultural issues, some of these problems do get wiped away when you virtualise things.  You can take away some of the problems.  But other ones, they just - they're really, really stick problems.  So attacking that margin is really challenging.  We add one more thing today, I've got my huge investment that I've just started making in Oracle UIM and that's going to be nicely integrated with me ordering system, with Siebel, but I have to integrate Siebel with the orchestration system that I already built and I'm going to replace with, and CENX is getting a little bit of visibility into some of our problems.  But how do we take our existing order flow, go through end-to-end activation, follow the [Tom] model.  Then what's SDN going to do?  What's NFV going to do?
With NFV - or with SDN and NFV, remember I'm going to take a lot of classical features like protection and I'm going to start charging for it.  I'm going to do the opposite of - from a billing perspective, I'm going to do the opposite of a mashup.  I want to start charging for features.  How much protection do you need?  Do you need 24 by seven protection?  Twenty-four hour by seven protection?  So today, our customers tend to get either protected or unprotected service and it's equal ratios, but that's not what they need and it's definitely not what they can afford to pay, nor can our margins handle it.  So if I can offer them protection as a service, peel out - but unfortunately, why do they want a cloud connectivity?  Because they have a VM here and a VM here and I need to connect them.
But Amazon and Google, when you buy VMs from them, they give you the connectivity.  They don't need a license from the ministry.  They don't need licensed telecom engineers.  They don't need to - bang, it's just - comes with it.  So we've got a tonne of problems as cloud connectivity providers that is really, really tough.  Google, Amazon, they don't need all of that NFV and all of that SDN. 

Phil Tilley
I would argue, actually, that I think Google and Amazon have actually built that in from day 1 and that's part of the challenge here; is Google and Amazon have started from greenfield, using IT cloud technology, [multiple speakers].

Gint Atkinson
I realise that.  I realise that.  You've got to remember also, think where Google and Amazon are built, also, where they're getting their 100 gig.  But the point that I want to focus on is, they cannot charge for connectivity.  They're not licensed.

Phil Tilley
Yep.

Gint Atkinson
So that simplifies their SDN and NFV and possibly allows them to focus on areas where they will get bigger bang for the buck instead of legacy functions that I, as a connectivity provider want to start charging for.  They just greenfield it, forget about it.  We don't even need it, we'll just bypass it.

Phil Tilley
Yeah, but I think you also have to look at that their business models are working on one operator system admin per 10,000 servers.  Extreme automation.  Yes, it's in a datacentre but that's where they're starting from. 

Gint Atkinson
And extreme simplification.

Phil Tilley
Yeah.  I think that's - but that's where the competitive pressure point is.  That's where the traditional service provider has to get to a same level of automation, same level of efficiencies where they're at.

Gint Atkinson
Unless - remember, they don't have the assets.  They can't trench - in Japan, they're not going to be able to trench the [fibre], they're not licensed.  So carriers will get forced back to being an asset-based business.  So if that's what happens, what does the industry structure look like?  That's - with the price points dropping down, the price of the bandwidth, eventually you're going to get down to all that's there is either the intelligence on the connectivity or all of the assets.  Because they can't get to the customers, they can't get to the mobile customers, they can't get to the mobile customers, they can't get to the [eyeballs], they can't get to the residential and they can't get to the enterprises without the physical assets that the carriers have.  But what does that economy look like?  What's the share of revenue?

Michael Howard
Ulrich?  Or any other comments here, James?

Ulrich Kohn
Yeah, I think it was quite surprising to understand Andrew talking about value of SDN, NFV, the challenges and the status of the technology.  I think there's a quite high alignment in the industry.  So what vendors are doing, if you look at NFV, that centralising of functionality.  You talk about a very simplistic [demarcation] device, cost optimisation.  There will be some secondary challenges.  So if you put functions central, there might be some interaction with the demarcation device which is not happening before.  So as you, intrusion detection, you want to do it centrally, you need to monitor all traffic which is de-central.  So with the rising probabilities - some selling or [following] tables, overflow comes into play.  It's a new feature, a new function.  We need to solve those ones and the industry [unclear].  So this alignment, I think, is quite a strong thing.  
SDN, I like your statement saying you would like to see that commercially available as soon as possible, you see the advantage of having bandwith, on demand services.  So we just put together a demonstration, together with British Telecom and we worked together with [Npower, some] work for British Telecom, we worked with solutions which are used [unclear] the [media] industry for resource management of those companies.  It was really excellent to see, in an open environment, in an open connectivity environment, the implementation of a self-provisioned solution.  It was a matter of hours rather than [maybe you have] [unclear] integration that [unclear] management systems and you [just start] to develop and implement those features you talk about months and almost years.
So I think we are now starting to realise the huge benefit of the [openness].  While I'm not sure whether we have fully understood what the impact on the networks will be.  I can give you one example.  Openness for me is very much a cost reduction in network integration.  [It's getting] [unclear] into standardised open interfaces.  If you go back some years then an argument of vendors has been, [a] lot of vertical integration in network elements for the reason that only have to do the integration once and then I can sell it to various customers and they don't need to do the integration [unclear].
So if we are now looking at an open environment where the integration effort is less extensive, less troublesome, less costly then the question is, [don't] we need to go back and challenge the question of, does it make sense to integrate vertically?  Wouldn't it be much better to have a best [in grid] selection of various technologies to take [unclear] switches from the enterprise datacentre environment in [unclear] network in the future.  It might be more cost efficient.  Combine it with [unclear] layers and the multi - based on [unclear] solutions, a more cost-efficient optimum architecture.  So [all I'll say is] probability on average of might change of layering architecture, so we should - and we will see a next step of innovation and thought about architectures in [unclear].

Michael Howard
You bring up - sorry.

Chris Purdy
I was going to say, the one thing that I find interesting first of all, the value of this to service providers.  One of the biggest - first things you actually have to do is what you mean by SDN and what you mean by NFV, what you mean by service orchestration.  Because what ends up happening is these terms have taken on a marketing perspective.  It's very hard to tie.  Because what ends up happening is, for example, they'll think of NFV and that gives you service flexibility which means people think, oh that means connectivity on demand.  But those two don't necessarily follow from each other.  Because even today's [MPLS] networks, if I want to increase the bandwidth of a service, I need to change the [unclear] at the edge.  The rest - the network takes care of everything else, even in [unclear] of [VPLS] service.

Gint Atkinson
This is GMPLS...

Chris Purdy
[Multiple speakers].  Same thing.  So we already have a lot of these capabilities but a lot of the barriers were in the management systems and the structure of those.  So if you include orchestration as part of your overall [SEM] world then you see some of those benefits.  So I think my first comment is that until - it's hard - you actually have to be very specific of what it is you're talking about when you talk of what the benefits are.  We don't tend to be.
The other comment that I'll make is that I find it quite interesting that NFV actually, I totally see the potential value of it.  The whole notion of this shared compute infrastructure and rather than installing a box every time I can install a virtual network function and instead of physically cabling it to other boxes, I can set up logical connections within that.  So I can totally see the value of that.  But I've heard about it for a long time and I go into service provider after service provider and I see activities going on in labs but real world implementations?  And similar on SDN controllers.  I am seeing, obviously, some deployments of orchestration systems, but it's...

Gint Atkinson
A point on that [would be] if there's real - core function, there's real traction...

Phil Tilley
I think that's exactly.  If you look at AAA as an - or DNS as an example, [multiple speakers].

Gint Atkinson
With those - those already are virtual - are running on shared computing and [multiple speakers] in the cloud [unclear].  We're also running our voice switches on NFV on Juniper [multiple speakers].

Chris Purdy
If you define it as running on shared computers [multiple speakers].

Gint Atkinson
This stuff, there's a lot of tangible stuff happening and firewalls and...

Phil Tilley
As virtual is, I think one of the things...

Gint Atkinson
But this is easy.  We did this four years ago in the datacentre, selling services to the financial industry.  We were the SDN controller, we wrote the orchestration engine and how many vendors - everybody.  I've invited everybody, bring in your SDN controller and just do the top 20 per cent of what our SDN controller's doing and you've got the business if you're not going to ask for a couple of million bucks.  Come in, bring in your orchestration engine, let's - and then we want - we don't want to reinvent the wheel but again, we're not even getting vendors to come into the lab.  We have a few, CENX is offering to come in and we want to rock and roll and do a POC.  We have a few vendors but many of our most important vendors that are advertising a tremendous amount, they're ready to take an order.  But they're not ready to go through [TAB] testing.

Michael Howard
So if you're thinking of the state of the market...

Gint Atkinson
They're not ready to go through TAB testing. 

Michael Howard
Right, the state of the market is still early and young.

Gint Atkinson
And commitment's two years old and the schedules are already two years out.  Sorry, we can't do it.  Okay, well I'll be real happy if AT&T does it and you guys make a bunch of money and groove it out.  I'm happy to wait a year or two, but that hasn't happened either.  So we're the young, little guy, asking for people to come in.  They're not showing up.

Chris Purdy
I think it's going to - I think what's going to drive that is - we are starting to see certain service providers that are starting to have better agility and are able to do it.  But until they are in there - because Google isn't a competitor providing connectivity services, right?

Gint Atkinson
Well - but indirectly they are.  The customer that's buying VMs and coming to me for connectivity from here, from this datacentre to that datacentre, they're going away.  Now I'm lucky that [- though, that the] top layers are coming in, buying capacity, but that's a radically - that's at a wholesale price point.  That is insanely different from selling this customer 500 meg to there.

Chris Purdy
Right.  But I think - hoping that in 2015, we start seeing more providers.  There are real world examples and I wouldn't [unclear] offer increase your bandwidth on demand by a website, a certain percentage.  So they've - they are [pre-engineering] in that work, they have got the flow through, they've got the orchestration in place, as an example.  I've seen datacentre operators [we're] orchestrated, CoreSite, for example.

Michael Howard
Well then the market's still there [multiple speakers].

Phil Tilley
There was a customer here in Germany five years ago, TeraStream, I think, launched a service five years ago with a portal to allow you to dial up the bandwidth.  That service never took off.

Chris Purdy
Yeah, so until enough of it takes off and it would force everybody to be as agile.  Today, if you were - tried to go in with an infrastructure as a service offer and you couldn't give compute on demand, it took you 90 days to turn it up, you couldn't be competitive.  Because that's the market norm.  Whereas connectivity services, the market norm is 60 to 90 days.  Until more providers are actually able to do it a lot faster, everybody's...

Michael Howard
We're getting out to the end of our time and we've almost finished one question. 

Andrew McFadzen
Maybe just one last comment though, just to this if I may?

Michael Howard
Okay.

Andrew McFadzen
Well, there's two parts to it.  One is - the good news is that we are doing proofs of concept, which is positive.  I think the other thing that we have to recognise and remember - and I'm talking here about the big telcos, the big service providers, is it's not only a question of adopting a technology, we've got to change the whole organisation.  At the end of the day, we're moving from network engineering to IT.  Culturally, educationally, skillset of the people, it's a massive shift.  Those shifts don't happen quickly in big companies.

Gint Atkinson
Yeah, the [carrier] ops guys, they culturally cannot handle the DevOps the NetOps model. 

Andrew McFadzen
Yeah, so - and that in itself...

Gint Atkinson
In general.

Andrew McFadzen
...will dictate how fast the market moves.  Okay, there are new market entrants and we have to be concerned about them.

James Armstrong
I would just add something here, that I think the potential for cost savings in the fixed costs are there for NFV, and I think that's very clear.  I think the flexibility holds a lot of promise.  I think the question is will it really work reliably to the extent that the enterprises have now drawn accustomed to the reliability of the physical network.  Can NFV at load, when it's fully adopted, when it's fully rolled out, will it work?  Then if there's a problem, the question is, how much extra cost is there to figure out what went wrong?  So is the cost of ownership going up on that side in order to be - to compensate for the flexibility and everything else that it offers?
So I think those are still out there as question. 

Andrew McFadzen
Yeah, you're absolutely right.  If a customer's physical firewall goes down, you can send an engineer out to fix it.  If we're running multiple customers on the shared - it's going through a firewall and that goes down, we're in big trouble.

Phil Tilley
I think that's the difference and this is where Gint's point is.  Actually, when you talk NFV, there is virtualisation, so I can run my firewall on a virtual machine, on a hypervisor, absolutely, and that's here today and [happily].  But actually, NFV is really - okay, so if that virtual machine goes down, how do I automatically set up another instance of that and scale up, scale out and redundancy and resiliency.  So I think that's clear when we talk about NFV and the definition of what NFV is, there is a difference between straight virtualisation and truth cloudification.

Gint Atkinson
Yeah, and then [don't] I want to charge for that and make it restoration as a service and charge for that [multiple speakers].

Prayson Pate
On that point, so they're - we can't fall in the trap of thinking that the only services we can virtualise are the ones we're offering today.  Your point about getting the revenue from different parts is very good.  There's also going to be an opportunity to distinguish between services.  So we all - most of us came out of the telco world where for years and years you had [unclear] quality in phone.  So who would go for a phone call that was crappy?  Well it turned out a lot of people would because it was free or almost no cost.  There was a whole different market for a different solution and a different customer set that led people to offer very low cost services as well as advance services.
I think that's what we're going to see here.  Your - I think your question's very good.  Can we initially offer the same level of service that people are used to?  And probably in a lot of cases, the answer's no.  But there's also going to be a lot of services that you can't offer today.

Gint Atkinson
Everyone's getting used to the voice quality online.  So when you land in Tokyo, it's really easy to get a data sim and phone and on the brochure they're showing you the phone you can rent, the data sim which doesn't require any identification and then a bit line icon and Skype and - so people are getting used to  a different experience.  But I think the problem is, the market timing, because the service providers and the platform providers have to sort out the roadmaps to intersect the demand and hit the revenue and margin.  The timing is really, really challenging.  The CapEx investment and the resource investment for development is seriously painful.  So you can see how quickly you run the numbers past the CFO and they say, the benefit is marginal.

Prayson Pate
Well that's my point is there's going to be applications where there is tangible benefit.  That's the ones we have to identify to drive these early innovators.

Gint Atkinson
We found a lot of them, too.  In the datacentre, it's easy.  Because the guys doing operations there are also culturally different.  But as soon as we get out into the WAN, the connectivity service, whoa, we've got these telco heads.

Phil Tilley
Also, I think in the datacentre as well, the pace of change - the pace of change in the IT community is much faster.  In the networking community, that natural pace of change is, hey, what are you do - you can only do a software upgrade historically on the network every two years.  So we generate - we'll produce a software release every year but it's every two years that that actually gets implemented.

Gint Atkinson
But isn't it amazing how many software upgrades we tolerate?  Not only in the apps, but over the air?  The firmware updates that are happening?  So it's interesting, why can't we move this culture in the network?  And why do these radio engineers expect - they accept software defined radios for over a decade now and over the air provisioning, firmware upgrades, everything.  But why can't we do it...

Prayson Pate
Because they've identified an application, people are willing to [multiple speakers].

Michael Howard
They will do it.  They will do it, it's a matter of time.  How many years [multiple speakers].

James Armstrong
That affects one user as opposed to millions.

Gint Atkinson
No it doesn't, it does not.  When I push an update on the firmware, year, at T-Mobile USA, believe me I know, I launched the first over the air upgrades all the way back in 2002 and then I launched the second web data service in the world at KDDI, [easy web] and yeah, that's - they're terrifying.  You're talking about 40,000 users, you should see what happens to the call volume, the inbound calls.  They're not going to - even today, they're calling on the phone, they're very - in that context, they're very unaccepting of going to the web-based help.  Whereas when they use their web-based service, they're willing to get web-based help.  But when you're using a mobile web service on their handset, they're going to call in. 

Michael Howard
So this has been fascinating, we want to have time for the press folks to ask their questions.  Just before we do that, I would like Marc Cohn to briefly talk about what does it mean to be open.  We've brought up open early and I wanted to ask you this.  I know you've presented this before in the past.  If you want to answer it.

Marc Cohn
No, I certainly want to answer it. 

Phil Tilley
Marc, tell us about open.

Marc Cohn
I think open is what SDN is all about.  I think of - when we talk about this discussion and we talk about what is the functionality, separating and controlling data.  Well isn't that the way [voice network] operates and the transport network operates and a lot of other network operates?  [Unclear] wireless LANs.  If we look at some of the new services bandwidth on demand, well that's not so new as well.  What's different is two things.  One is the programmability that we did talk about.  Being the ability to change the network behaviour which translates directly into differentiation, which I think is of interest to the operators because of the ties to the vendors before.  Then also the innovation that new services can now be whatever we want them to be or they can be dynamic.  We don't know, we don't have the [unclear].
But what's really - what's unfolding with SDN and NFV and as a participant in both watching it unfold, it's this challenge of how we bring this - how do we create  new platform that allows multiple participants to be able to thrive together instead of trying to create [their own] individual platform dominated [unclear] limit how others can participate.  There's two ways to do that.  One, I think, is open source, and Neela's on the other side of the table with OpenDaylight, platforms like that that are not controlled by any individual organisation.  The second is to - is for operators to take their newfound empowerment and then be able to insist that multi-vendor is not just the challenge, it's about making it vendor neutral.  Which is really what openness is about.
If not one vendor can control the infrastructure, that's open, that's truly open.  That means everybody has to be aware and everyone has to converge on the direction that operators want to move into. 

Gint Atkinson
I'm ready to place an order.

Phil Tilley
No is that a product - it's open, so is that a product or a service order? 

Andrew McFadzen
[Multiple speakers] strangely enough [multiple speakers] collectively the service providers will be able to influence that leverage but we don't seem to be able to.

Marc Cohn
Well I think you are.  I think that's what NFV is about, that's what the open source movement is about.  I think it's because operators are empowered and they're forcing us into the hand of [unclear].

Phil Tilley
But I think the importance of that, though, is for an open source project to really take off and be successful, the service product also has to put some skin in the game.  So they're taking that open source.  You can't just take and take, there has to be some putting back.  so I think the open source projects that are really successful is where the service providers are also contributing and having that development community, which isn't just vendor-led development community but service provider.

Marc Cohn
It's evolving already, Phil, with the open platform for NFV which is built on other open source projects where the operators are taking a very active role.  So I think it is happening but we're [multiple speakers].

Gint Atkinson
But the reality is, the customer experience is this.  The vendor product teams, the habitual response of product management is, why are we in [unclear] competition and you guys definitely, you know that you always hear that in the product strategy discussions within your companies.  It just - it's not happening.  You show up and you do these [enter] ops and then you can't show up in our labs and do the enter ops.  You go to enter op world but then you come to our labs and it can't pass the test, it doesn't work.  So - and on top of it, can't even get you guys to show up and do a POC, it's really, really tough.  Now it's starting to come together, see the product launches and it's - maybe this year, we're going to be able to get rid of all the stuff that we had to build ourselves.

Chris Purdy
I think a lot of it comes down, and a part of the challenge of doing a POC is that there's a cost to the vendor to come in and do the integration with the [unclear].  We did one POC where we integrated both into a bunch of existing operational support systems and element management systems.  You're talking about [unclear] integration, TMF 814 and all this old stuff.  Then we actually had to integrate into a [CNS] SDN controller sitting on top of the orchestration layer.  The difference was - in this case, honestly, it was night and day.  The challenge we faced in integrating the old stuff was probably - I don't know, 30 times the work that it took us to integrate into the SDN controller.  We got the documentation, it was done, it was open, here it is.  It was as designed, et cetera.  But that is not the norm in all case. 
So what ends up happening is - your challenge whenever you come in to do one of these POCs, because every operator has an environment that's sufficiently different, there's a lot of old stuff that needs to happen to integrate into, plus the new stuff in order to get the value.  So that's where the - and so all vendors are going to struggle with that [multiple speakers].

Gint Atkinson
But remember you're asking...

Marc Cohn
[You're struggling] because you have to provide the migration [path to me].  I don't think any - we haven't talked to make operators, I can - other than some alternative operators who have the greenfield luxury of being able to just ignore the quote, unquote, old stuff which we spend a disproportionate amount of time trying to integrate with so we can migrate to the vision of SDN and NFV.  That's not going to go away [multiple speakers].

Gint Atkinson
Okay.  So you're just telling me I can't get my time in the market and I've got to wait for you guys.  So...

Marc Cohn
No, I'm saying just do it greenfield and it's a lot faster.  [Multiple speakers]  I think there's a fallacy that [multiple speakers].

Gint Atkinson
[Multiple speakers] and telling us, oh you're going to be able to launch those services so fast.  Not even including the integration work.  Oh you still have - no, that's not working yet.  No, we can't - we'll do the restoration, we can't do this, we can't do that, and here's the roadmap.  It's not going to...

Neela Jacques
This is the challenge with the proprietary work and I think...

Gint Atkinson
Already have it working.

Neela Jacques
I think people have been talking about this question of, okay, well is it open source or is it, do I go proprietary vendor?  And I actually think that the world is changing.  Fifteen years ago, you were talking about [Richard Stalton] versus Bill Gates, and that was the battle.  But that's not the way that it is right now.  What we hear over and over again is people want open source and vendor support [multiple speakers].

Gint Atkinson
All I need is a plug-in for OpenStack and for Daylight for one service.  So every service, I need a plug-in, I'm done.  It's about 100 lines of code and an object model that's already defined...

Neela Jacques
And you might be...

Gint Atkinson
Why should I pay a lot of money to develop that small piece of code?

Neela Jacques
You might be - and I would say, what I see around is there's a range of people with ranges of capability.  On one extreme, you've got Twitter and Google and Facebook who are like, I don't want to pay anybody for anything, I've got all these PhDs, I can invent it myself.  On the other hand, I've got a number of operators in some developing countries who are like, I have almost no capabilities, just give me something that works.  We've got the full range, certainly in the teclo operator side, we see the same thing in the datacentre.  So back to your point around, well, if someone wants to come do a POC, but what if they're not ready?  I think that we're moving into a world in which open is a way of doing, in a sense, the basic research.
It's allowing us to fast-track the standards process.  To innovate but get to de facto standards and provide a layer of technology which anybody can access and anybody can use.  Then from that, there's a range of offerings that get built up where proprietary adds the differentiation.  Some people will do a light layer around it, some people will add a full, end-to-end solution.  Some people say, I've got to have it perfect and it has to be tested to the Nth degree.  Great, someone's working on that for you on top of open source.

Phil Tilley
That's great in open source if that development community, contribution community is large enough and the innovation is happening.

Neela Jacques
Absolutely.

Phil Tilley
But actually, if you say, okay, well here's a piece of open source code and the only person contributing is one vendor, it suddenly becomes proprietary and not open any more.

Gint Atkinson
Well yeah, that is a problem.

Neela Jacques
Sure.

Phil Tilley
But I think...

Neela Jacques
We've got a few of those, single vendor open source projects and, hey, it's open, because I threw it up on GitHub.  The value of an open source project has to do with the size and scope of its community.

Phil Tilley
Exactly.  I think that's where OpenStack is certainly...

Gint Atkinson
But once they get scale, one it gets the right size, the open model's going way faster than the traditional standards model like with TMF, with IEEE, IETF, carefully engineered, orchestrated agreement.  So the open source, once there's scale, once it gets to a letter, is moving faster.

Phil Tilley
It moves very fast.  I think - and that's where...

Gint Atkinson
But people have to contribute.  KVH, we're going to be putting a whole bunch of contributions out into open source.  We're talking to vendors like CENX, well shouldn't we take these set of requirements and really make this available to the whole industry and put it open source under OpenStack?

Phil Tilley
I think that's where open source becomes valid.

Neela Jacques
Yeah.  If you look at OpenDaylight, it's April 2013 and [unclear], we had 2.5 million lines of code in OpenDaylight by the second release, with 280 developers with a multiplier factor of anybody really doing anything with OpenDaylight has 10 times as many people contributing the project, is my guess.  It's maybe 10 or maybe higher.  That's a significant critical mass that any competitor is going to be able to address.  It's a very compelling path to openness, but it's a journey not a [step function] that all of a sudden we're here and then the open [unclear] system [unclear] and we have [unclear].

Michael Howard
I like the notion of external R&D.  So you have someone developing someone and if you have a community of open source, it's like an external R&D that you're adding.  But you're - actually it turns around, you're contributing to the whole.  So let's open the floor to - there must be some questions that you had on your mind before you walked in the door and more since you've been here.  Who'd like to start?  [Ralph], [Barry], Carlos?

From the floor
From my point of view, I have seen that all these [third] network issue mainly focus on reduced cost.  I don't feel that this change will increase the revenues of the [admin] service provider.  So maybe...

Michael Howard
Let me give you an example what Telefonica is doing and what they'd done with their - a small part of their consumer base in Brazil, they implemented what's called the CPE, virtual CPE.  They're offering IPTV services over that.  What they did was select a group of their broadband customers that had a router and now that router by Telefonica is treated as a connectivity point.  What they have now on services running in their network is a whole range of IPTV services that just weren't available before, period.  They have firewalls and they're talking about having three grades of firewall. 
In the past, what they had to do was it took a year and a half to say, we're going to offer a new consumer service that has a firewall with it in our broadband and they had to find the service, find a vendor that has all the characteristics they want and a piece of hardware, test it all out in a lab then roll it out and then finally get money back from their consumers.  But in this case, now that they have everything set up, they can, even though they're offering one firewall, they can now offer three grades of firewall and offer parental control. 
It's all software so rather than - so they're able to get some little incremental piece of the revenue off of their...

Phil Tilley
But I argue - yeah, so I argue with that, actually, the amount I pay for that service, I'm only paying for so much communication or entertainment.  Me, as a personal user, pays so much.  So much of my income goes towards communication and entertainment.  My budget that I'm prepared to spend only so big.  So Telefonica, great, you're giving me more stuff, excellent, I'm not going to pay you more.  But what I will do is I will stay with Telefonica if you offer me more.  So I think that's the - it's about the customer retention.

Gint Atkinson
Yeah, in Japan...

Marc Cohn
Cost reduction.

Gint Atkinson
...my residential service, one gig, and 100 meg is always nailed up, flawless for internet browsing and so forth.  But every - all of the content, flawless.  So I get one gig, I also get back to - I'm grafted into an ELAN instance, get back to the office.  But I've got one gig and I can fire up lots of apps.  But it throttles at about 100 meg per stream.  But if I fire up stuff, it's all going great.  Seventy bucks a month, seventy bucks a month, and I get 200 IPTV channels.  Of course that's only SD not HD resolution.  But I got the bandwidth for it.  That same service when I was in the US, I was burning through 400 bucks a month but I was only getting 100 meg service.  So I think a lot of the world, the pricing on all of this stuff is dropping through the floor. 
So the new revenue, if we say increased revenue, I think it's challenging.  But if we say new revenue from another service or from a service mix, definitely it's there.  But unfortunately, the effect may be that it's just providing [unclear] for...

Michael Howard
Yeah, but won't you pay - don't you have the [yen] to pay for another buck a month for your teenagers to put parental control in?

Gint Atkinson
Yeah, that's the idea, is to sell more features or more packages, more bundles.

Andrew McFadzen
Just coming to the question.  I think maybe - and again, I'm looking at it from an Orange perspective - I think one of the reasons why we look at it from a cost standpoint, cost saving standpoint first and foremost, is that like all of these technologies, the people in our organisation that are looking at it first are the engineers and operations guys.  They tend to look at it in terms of operational efficiency or engineering advance, which is equated to cost.  They don't think of it in terms of marketing and product opportunity. 
Which is why, I think, you'll hear the service providers by and large talking about it as an opportunity to save cost.  I think we're getting to the point now, though, where we're starting to talk to people and we're starting to talk about incremental revenue opportunities.  Putting aside the ability to deliver new services, and I think absolutely we will be able to,  but just think of the increased revenue potential we would gain if we do - turn up a new service in one day rather than 90 days.  We gain 89 days' revenue per connection. 
If we can bring a new product to market, today it probably takes us 12 to 18 months to develop and bring a product to market.  If we could bring a new product to market in a month or faster, which is what happens in the IT world, in the software world, big revenue potential.  As I say, I think we're - we're getting to that point in time where we're moving from saying, well how much is it going  to help us cut cost to - okay, well let's do that but let's also look at how can we do other services off the back of it and revenue opportunity.

Neela Jacques
I wanted to just - a point on - Phil, you said this, you've only got a certain budget yet in telco that's a debatable point.  If we look at the broader picture, how many of us five years ago had a phone and were willing to pay over $300 for a phone?  Almost none of us.  Yet every single one of us is sitting here right now with a $600, $700 device on us.  It's because at some point you look at it and it was - if you take the stupid user on it, the value of being able to check into your airline, the value of being able to watch movies on the go, has meant that we are paying a lot more.
Now, it happens to be that the innovation has been primarily captured by the hardware device maker and unfortunately from carrier perspective, the carriers are - have been stuck thus far in a model where they haven't been able to monetise on innovations to be able to capture that.  but I think the bigger picture still is, if we look at IT's percentage of the value that's going on in our society, if we look at the value of bandwidth, if we look at things like internet of things, I think there is tremendous potential for being able to actually extract extra dollars from end users.  Part of the challenge is we've been stuck in a model where it's so difficult to be able to innovate and offer something new.
But the moment things become relatively easy - I know, I bought 800 megs for this travel here and it cost me an extra $120 a month.  Frankly, if they'd offered my 1.6 gig, I'd probably pay that too.  I think part of it is we're stuck in a model where we've been hardware, hardware, hardware and we've been duct tape and paper clip for so long.  I personally believe that as we move towards SDN and NFV, we will be able to increase the agility and the cost of being able to change something is going to go down to the point where suddenly we're going to unlock a whole set of use [cases] that we don't know now.

Phil Tilley
I agree.  But I think it - my argument with this is, yeah, you're paying more for this device - actually, you're not, because this device in most cases is subsidised by the service provider and you will go to the best offer for the different service provider.  So compared to the [unclear] actually the price doesn't go up much.  But the one that can get you those features faster, that can offer the best deal, is the one you go to.  So the choice between carriers is opening up.  That's where NFV and SDN becomes critical because as competitive pressure says, actually, I want to go to the one that can offer me the service in one day, not the one that can offer me 90 days.
So I still think the new revenue comes from actually customer retention.  Customer acquisition and customer retention.

Michael Howard
With that, thank you all for participating in the panel.  Thanks for all of the wonderful questions that we got from the press.  Sorry about the time.  But I think we're going to be around here for a little bit if you want to have a one-on-one question [unclear].  But thank you, panel.

 
Above, PHOTO / telecomkh.com 
valorar este articulo:
Your rating: None

Post new comment

Datos Comentario
The content of this field is kept private and will not be shown publicly.
Datos Comentario
Datos Comentario
Submit