Orange’s Director of Technical and Network Strategy is yet to be convinced by some of the key mobile networking trends that were exhibited and promoted at Mobile World Congress.
Yves Bellego, speaking to TMN, ran through how the operator is assessing and developing technologies such as NFV, edge computing, small cells and VoLTE. On NFV Bellego said that there are still pieces of the architecture and vendor overlap that need to be worked out. “Every vendor says they can do everything” he says. He’s not (yet) a mobile edge computing advocate, and neither is he convinced small cells have a role to play any time soon, beyond “special projects”.
However, on another hot topic – WiFi-cellular aggregation via LAA/LTE-U he was more bullish.
If you are seeking an update into the operator “mindset” I believe this is a revealing interview. It also works as a nice update to the interview Bellego gave last year, in which he addressed many of the same topics. Back then, Bellego said that Orange would be taking an incremental approach to many of these technologies, almost cautious. He was also glad that there had been so little hype of 5G (in 2014)!
So, as virtualisation, the cloud and 5G ran rampant at MWC15, TMN asked if his views have changed, and what Orange been doing in the interim on these topics.
Cloud, NFV and virtualisation:
Last year you talked about NFV, SDN and said your view then was that you would go slowly, implement one instance and test that against one hardware item. Then perhaps bring another vendor in and test again. Then you might expand. Did you do it, what happened, where are you now?
We still believe that NFV will not come in one shot but in various pieces. What we did in 2014 was a first Proof of Concept (and we made an announcement and demo on the HP booth) of some trials of vEPC, and vIMS with HP and Nokia. We tested the orchestrator, virtualised functions, so we are kind of on track.
One thing where I do not see as an industry that we are making huge progress is the overall architecture: we will need orchestrators, VNF managers, so what is the exact function of each of these, how that fits together, all that is not clear today.
Today if we put it together there is some vendor overlap. They [vendors] all claim to do the same thing, so at the end where do we do that type of function? We need to ensure there are not some missing elements, especially in the management of VNFs.
So the ETSI MANO framework is not doing that for you?
That is good but probably not sufficient so we have to work further on that.
At the end what we need is an NFV system that is open so we can really use an orchestrator from one vendor with different functions from different suppliers, and hardware all from different suppliers.
Our questions are, what is in the orchestrator, what is in that layer, what is in the VNF manager, how is all that is put together? Today if we put it together there is some vendor overlap. They [vendors] all claim to do the same thing, so at the end where do we do that type of function? We need to ensure there are not some missing elements, especially in the management of VNFs.
What is the goal and benefit of virtualisation for you?
When it comes to the service platform that is already quite a lot virtualised, quite a lot in the cloud. For network elements, the initial benefit is scalability and agility – and then we expect some opex savings. But that requires us to review our processes which is not easy. We do not really count on capex savings, that is not an issue. Already today we are on quite standard hardware and all the vendors claim they are already kind of virtualised – but in their own world.
You mentioned last year the organisational challenges of bringing IT together with the telco networks operational side. Has anything happened in the business over the past year on this?
Not so much. This is something… we need to prepare the operational teams but they are not yet today to really work on virtualised functions. This is new for the coming year, and is today more a topic for our suppliers. Our research teams and our what we call anticipation teams that were initially network teams have trained on IT systems. But for operational teams that will not come in 2015. We have a bit of time to prepare for that.
We really believe we will need some virtualisation in the network points of presence… we are not yet really looking at it on the base station
One area of progress where we have a clear view is that we do not believe cloud or virtual extensions will be in just one layer of the datacentre. We really believe we will need some virtualisation in the network points of presence. That’s the thing.
We see ideas similar with Nokia’s Liquid Apps, we also believe there will be not just one single layer of cloud, we will have different layers, orchestrators that manage both centralised and more local instances and functions. This is what we are looking at.
You reference Liquid Apps – that takes processing right out to the base station…
So in that project we work with the cloud in Network PoPs, we are not yet really looking at it on the base station. What Nokia is doing with Liquid Apps – in terms of the product it works well, but whether we really need that and when is not clear for us. We think there may be a need for storage, processing but on network pops – not on radio side – more on a metro or regional level.
Last year you said how glad you were how little conversation there had been on 5G. Take a look this year…!
What we did in 2014… we did all agree that 5G needs to be prepared now and be available by 2020. What we did not believe last year is when people were saying it would be commercial in 2018. But today you would not hear anyone say from the suppliers that would have commercial 5G by then. [ed- oh yeah?] Today it is in research, standardisation is coming soon: in 3GPP there’s no work item on 5G today. And we have agreed with NGMN that we will do step-by-step what we want, develop what we need and we will get a real advantage in 5G.
So maybe the requirements are not that clear today and 100% what we need but at least we have some guidance and real improvement compared to 4G. We really need systems that are quite flexible, because we don’t know the needs of 2025 but we do know they will be quite different to the needs of today.
Small Cells, indoors and network densification
So let’s get more tactical about the network, and talk about network densification and the possibility for differentiation around indoor coverage. Last year your said you would fix macro first, then go denser where required, and work from outside to in…
This is what I planned to say today! We first need to reach 95% coverage like 3G, do a bit of small cells for special events, at still a limited number of sites. Small cells are all beyond 2015, WiFi, femto, pico, Dot – whatever – it is is only special cases. Large deployments will not be before we have solved overall coverage.
If you look at the vendor side, Huawei’s Lampsite won a GSMA award, you mention DOT, Spidercloud is announcing deals with Verizon and Cisco… the vendors certainly believe there’s a real focus on quality of indoor capacity and coverage. Is that specific to certain markets or spectrum holding, your view is outside-in , big-to-small.
Our focus is on quality and as such that is an issue of indoor but the first answer for indoor coverage is through the macro site, because that’s the easiest, fastest and lower cost. When we have fixed macro sites – and with 800MHz have basic indoor coverage – then at some point in time we will complement that with capacity.
Here’s something new that was not talked about so much last year – WiFi and cellular aggregation as in LAA, LTE-U. In terms of hype that’s all happened in a year. Are you that spectrum constrained that you want to go and listen to 5GHz and grab some of that down towards the device?
The good news is that the battle between licensed and unlicensed is over. Operators, including a lot of people inside Orange, were claiming they would never trust unlicensed. Now they understand that although basic coverage cannot be provided by WiFI we can provide good service through WiF/unlicensed if we can get simple and seamless coverage. What we need is both the devices and the network to have that and that is coming step by step.
As an operator we need to make sure customers get good performance and do not have to switch on or off the WiFi, so it needs to be native in the devices and the network – that’s the best.
(Editor’s note: reading this back, this is not really an answer on LAA – it seems to be more referencing NGH and Passpoint capabilities)
So our last topic is VoLTE…something a lot of operators have admitted has been hard work
I will just confirm we are doing it right now, starting in some countries then all over Europe. Yes it is complex because of handover to 2G 3G, the impact on the RAN, core etc and everyone has to do a lot of integration. One thing that is not that clear is to get devices that are native VoLTE operating on all network operators.
Orange has no public date on VoLTE in France but it is coming. On the network side we are deploying – it’s more of a marketing decision on devices. One comment I will make – technically side we have one single project for VoLTE and VoWiFi. We want the network capable of deploying both services and then it is up to marketing how and when to launch.