Quick post on some thoughts from Day 1 here in Barcelona:
As predicted, offload and data traffic management are top-of mind. Too much cheap traffic transiting expensive networks. Yes, people are paying for mobile broadband on laptops & smartphones, but shoehorning gigabytes of Internet-destined data across a network designed for in-house, data-sipping operator-controlled apps is clearly a recipe for disaster. What's striking, though, is that the "offload industry" hasn't yet mapped the various solutions onto a matrix of use cases and deployment scenarios. I think there needs to be a lot more thought here - how do you selectively offload certain applications, but route others through the core?
In particular, a lot of the network vendors don't quite understand what "applications" are on the network. Many will tell you that "video" is an application. It's not, especially from the user's perspective, which is where any policy or tariff changes will bite. There's a huge *perceptual* difference between watching a YouTube video in large-screen mode in a browser, versus watching the same YouTube clip embedded in your Facebook page, where it's been shared with you by a friend. Yet the network is unlikely to be able to spot the difference. One is "media consumption", the other is "social networking and recommendation", and I think that in users' eyes, they are very different. How you capture this behavioural component (layer 9?) is going to be a difficult challenge.
NSN had an interesting pitch yesterday - creating a smartphone-optimised network. This is pretty much a first, as normally the network vendors ignore the role and usage modes of devices in the real world, and vice-versa. I spend a lot of time pointing out to network vendors that innovations in infrastructure technology always have device-side implications (hence my work on IMS-capable handsets, femto-optimised devices and so on). NSN observed that although laptops create most of the data traffic on 3G networks, it is smartphones that are much more aggressive when it comes to signalling traffic, and that in some cases this is to blame in clogging the network. They're working on a 3GPP mechanism for reducing the signalling overload, called Paging. I'll delve more deeply into it on my return, but it sounds interesting.
One notable theme here seems to be that LTE is being viewed with more reality - it's conspicuous that both NSN and Ericsson are touting go-faster HSPA+ quite loudly, as it's becoming clear that many operators are going that route, especially in Europe. I have my suspicions that many might skip "beta" LTE (aka 3GPP Releases 8 & 9) and wait for the bug fixes to appear in LTE Advanced (Release 10). After all, the first versions of WCDMA were pretty useless until HSDPA arrived.
Add in the issues concerning LTE spectrum availability, devices, indoor coverage, voice & SMS, roaming, handover, power consumption and the lack of a new business case, and I think that we're looking at 2014-2015 for any real mass uptake. Apparently there's a few early handsets on the horizon, but I suspect they'll be provided with a free oven glove as an accessory.
During today I'm going to brave the dangers and gauge the state of the mobile IMS zombie. Yes, it's dead but still moving - but is it like the shambling creatures from Dawn of the Dead, or one of the scary, fast-moving vermin from 28 Days Later or I am Legend? The big question has to be whether it's already bitten LTE so hard it won't let go, and infected it. Anyway, I'm seeing the GSMA, the NGMN and various vendors with IMS-based products, so I'll try to avoid getting my brain washed (or, indeed, eaten).
Dean Bubley's Disruptive Wireless: Thought-leading wireless industry analysis
27 comments:
The Qualcomm booth over there at MWC should have something on offload exactly along the lines of what you are talking about (3G/WiFi interworking, offload, mobility, etc etc)
Where's the IMS bashing article in light of the GSMA adoption of OneVoice/IMS for LTE voice?
IMS-bashing to come, a bit too busy today doing real work rather than writing blog posts....
... but in a nutshell, the GSMA announcement looks like it has two groups of companies:
Group A are desperate for IMS
Group B are "yeah, whatever, if it happens then great, if not then so what, we might as well sign as it wins us brownie points"
You never disappoint DB!
Of course Group A have to be "desperate" for IMS, it couldn't simply be a more pragmatic "see IMS as a reasonable solution to the question of what technology to use for VoLTE."
Surely Group B are not merely saying "whatever?" Are they not crying into their beer lamenting what could and should have been?
I think that group A have an awful lot invested in IMS (either as vendors or operators) and don't particularly want to write it off, and start again.
IMS is such a "religious" debate that your phrase "reasonable solution" rather ignores the political and philosophical dimensions around it.
It's notable that the GSMA / NGMN stance is that IMS is rather nebulously positioned as a "target", with some of the participants having hoped to insert the words "long term" or "eventual".
As I pointed out at the time of the original OneVoice announcement, I am (broadly) positive about it at one conceptual level. I have long said that it's essential to have a no-frills, "CS equivalent", non-multimedia way of doing mobile operator VoIP. Doing a "bare minimum" IMS implementation is certainly better than full-fat MMtel + RCS + etc with all sorts of added fripperies like video and presence. I definitely believe that decoupling VoIP and RCS is a very good idea.
It's also interesting to note the Verizon / Skype deal. It maps onto a suggestion I made in 2007 that operators should gain experience in managing mobile VoIP (especially radio engineers) as early as possible. Simultaneously deploying a new radio network (LTE) and transitioning the main revenue-earning service to VoIP, with all the inter-dependencies, is a recipe for disaster. So "practising" with Skype on 3.5G makes a lot of sense to me.
And of course, it's establishing a Plan B hedge if VoLTE doesn't deliver.
IMS is only a "religious" debate to those who don't understand its weaknesses and strengths.
A lot of people have invested a lot of time and money in IMS precisely because it attempts to answer questions that need to be answered one way or another as the telecom world moves to IP. In its attempt to be all things to all operators it has naturally failed in some areas while at the same time being strong in others.
The fixed world is behind IMS, the endorsement by the GSMA for VoLTE is a very positive sign that the mobile world also sees IMS as a "reasonable solution." It's interesting to read your spin on this but your "walking dead" analogy is looking ever weaker and weaker.
What is your opinion on Nokia announcing the new MeeGo platform?
Davide - to be honest I haven't really had a chance to look into MeeGo.
However, I've never really been convinced by Nokia's attempts to call smartphones "mobile computers", nor Intel's attempts to define MIDs (and hence Moblin) as a massmarket device category.
I see Microsoft holding onto the computing platform crown for a long time, on laptops and netbooks. Apple's apparent slip-up with the questionable iPad seems to open a door for a #2 in smaller computing form-factors, which could be MeeGo or ChromeOS/Android.
I'm very unconvinced by the operator-controlled netbook/Smartbook rhetoric though. I cannot understand why anyone would relinquish control over key aspects of their computing experience to an operator. (I can obviously understand why an operator would like to *take* more conttol though)
I don't understand Nokia announcing the new MeeGo platform?
Access SBCs with P-CSCF providing critical control functions for VoLTE delivery security, service reach maximization, SLA assurance, revenue & cost optimization and regulatory compliance
SBCs and SRPs routing the SIP calls throughout the core network
Interconnect SBCs managing the off-net origination and termination at IP interconnect borders (as VoLTE starts as SIP, it will leave the network as SIP)
Wrt to the comments from Anonymous: Lots of things are "reasonable" solutions to problems. VoIP via basic SIP is also a "reasonable" solution for LTE. And there are many other "reasonable" solutions. Each of which has a cost. Of course if you are an IMS vendor (either base infrastructure or one of the numerous related equipment vendors) then you will no doubt want to charge an arm and a leg for your "carrier grade" network equipment so will tout the benefit of that. If you are in the core networks department of an operator you are also like to tout IMS as you think that some how it saves you. So anonymous you are right IMS may be less a religious debate and more a axe grinding debate.
"The fixed world is behind IMS ..." is a interesting comment. There are for sure several VoIP on the backbone with SIP deployments. There are also companies that claim that they have deployed IMS in droves. The reality is a bit different.
IMS is a very large beast. The 3GPP specifications are incomplete and moreover the set of specifications grows larger with every release. In reality very little of the specs are needed. Then there is the issue of how long it takes to implement and get it working for lots of consumers.
Dean is right the "onevoice" PR is not too different from many other PR's: there is a certain amount of "free PR", there are the vested interests people and there are the "We have nothing to lose" lot.
Bottom line is the GSMA adoption of onevoice is largely meaningless.
Original Anonymous here.
Can you name a "VoIP via basic SIP" product that addresses all of the scalability and regulatory problems faced by any large operator? If you can, it will presumably be implemented in some entirely non-standard fashion which will make for headaches later on.
While I don't think that standards are the be-all and end-all and I certainly agree with your comments regarding IMS standardisation, I contend that there really is no viable alternative to IMS at the moment (unless you are just building something in your bedroom to provide services to your mates).
IMS is a very large beast precisely because operator networks have complex requirements.
"Bottom line is the GSMA adoption of onevoice is largely meaningless."
Perhaps, let's wait and see will we? If they had endorsed VoLGA instead would that have been largely meaningless too or would you have seized upon it as another nail in IMS' coffin?
Original Anonymous:
"IMS is a very large beast precisely because operator networks have complex requirements."
That's only partly true. I certainly accept that large operators need very robust and scalable telephony systems. That's why fixed-IMS is making headway for PSTN replacement.
But what the operators *don't* need is all the machinery and compromises behind the letter "M" in the middle of IMS.
VoLTE, in my opinion, goes some way to rectifying the situation, by simplifying IMS as far as possible for mobile. But it still can't fix the basic problem that while SIP is fine for VoIP, it appears to be totally wrong for most of the other types of "integrated services" that actually give a decent user experience or business model.
3GPP is in a hole here - it can't really just scrap the whole thing as unfit-for-purpose, and reinvent some new web/developer-friendly alternative this late in the day.
Instead, it has to teach the elephant to dance.
The GSMA has given it some nice simple ballet shoes (VoLTE) and a rather ugly tutu (RCS), but there's no disguising it's still an elephant and not a ballerina.
"The GSMA has given it some nice simple ballet shoes (VoLTE) and a rather ugly tutu (RCS), but there's no disguising it's still an elephant and not a ballerina."
Fair enough, I think we can argue about how ugly the tutu is but the elephant still dances.
IMS does not mandate the use of SIP for every service interaction. There is no reason that an application could not use SIP based access for some aspects of a service while also using a more appropriate protocol for another aspect of the service and yet still have everything coordinated by an IMS AS.
"3GPP is in a hole here - it can't really just scrap the whole thing as unfit-for-purpose, and reinvent some new web/developer-friendly alternative this late in the day."
You are making the assumption that the whole of the IMS is "unfit-for-purpose" which is hardly the case. There is no reason that IMS features cannot be exposed in a developer friendly way either - granted, the IMS itself does not provide this directly but there is a whole slew of technologies and products designed to do just this.
Anonymous - yes, IMS can *theoretically* involve external developers. I remember going to a developer workshop Ericsson organised at VON about 4 years ago.
If it was that easy, maybe we would have seen some innovative IMS services with real appeal by now. Rather than pointless things like realtime video-sharing and SMS-on-IPTV and endless warmed-over operator versions of 2006's web applications.
And yes it *can* support some level of web integration, with the use of "a slew of technologies". On top of the existing slew.
The bottom line is that there's be more value (for developers, users and maybe operators) in creating "IMS for Facebook" rather than "Facebook for IMS".
The fact that nobody even thought upfront, that presence pings might kill the battery on a phone, demonstrates the lack of joined-up thinking between the core, radio and device standards groups. There's also a huge tendency to hide behind anti-trust whenever it comes to talking about user experience of business models.
Honestly, when I started asking people about IMS-capable phones and the UI in 2005, I just got blank stares - it was completely "out of scope". Same deal last year when I asked a senior 3GPP person (who ought to have thought through these things) about how IMS might work in scenarios with MVNOs. Blank look - clearly the first time he'd heard the two acronyms in the same sentence. Bang goes another business model....
By the way, I realise that I've mixed my mobile IMS metaphors horribly - dancing elephant, zombie, dead parrot and so on. I suppose I should have added "lipstick on a pig" for the complete metaphorical menagery.
"If it was that easy, maybe we would have seen some innovative IMS services with real appeal by now. Rather than pointless things like realtime video-sharing and SMS-on-IPTV and endless warmed-over operator versions of 2006's web applications."
It is that easy and clearly the reason you haven't seen it happen is that IMS networks haven't rolled out to any great extent yet. Take a look at what the JSLEE guys are up to for examples of what is coming. The fact that there are several approaches to opening IMS networks underlines the simplicity of the SIP AS model, you don't seem to appreciate that this abstraction will genuinely make it much easier to open services to third-party developers.
> Take a look at what the JSLEE guys are up to for examples of what is coming. The fact that there are several approaches to opening IMS networks underlines the simplicity of the SIP AS model, you don't seem to appreciate that this abstraction will genuinely make it much easier to open services to third-party developers.
As far as I understand, it's not really services that developers want access to, it's underlying capabilities of both network *and* device. Maybe I'm missing something, but I can't see how JSLEE and a SIP AS would enable a new 3rd-party IMS device-side client app (let's say a new form of messaging or social networking app). Would the architecture you envisage enable this to be pushed down to an existing handset, be guaranteed to interoperate with the other IMS apps and non-IMS apps on the device and so forth?
I'd expect 80% of the value in any new IMS service to reside in the UI and device-side software, and the remainder to be "in the cloud", yet none of the IMS client framework implementations seem to have a proper SDK to enable new plug-ins or extra apps.
A simple example: in an IMS context, how could a 3rd party developer write an application which intercepted inbound voice calls to say "the phone you are calling is on charge, and in a darkened room, in a timezone +8hours. Would you like to go straight to voicemail?"
"A simple example: in an IMS context, how could a 3rd party developer write an application which intercepted inbound voice calls to say "the phone you are calling is on charge, and in a darkened room, in a timezone +8hours. Would you like to go straight to voicemail?"
Why do you think this is not straightforward? There is nothing in your example that could not be easily handled in an IMS context. Clearly you believe that something in your example precludes IMS in some way but I don't see it.
"As far as I understand, it's not really services that developers want access to, it's underlying capabilities of both network *and* device. "
Quite right - should have said capabilities rather than services (although services could be similarly exposed of course).
Hello "Original Anonymous" You said:
"Can you name a "VoIP via basic SIP" product that addresses all of the scalability and regulatory problems faced by any large operator? If you can, it will presumably be implemented in some entirely non-standard fashion which will make for headaches later on."
I don't want to get into raging flame wars about soln 1 vs. soln 2. I agree with you that there aome "non-standard" Sip solutions there (even though that sounds like a oxymoron). I contend that we will have "scale" issues with IMS *both* technical and commercial. To my knowledgeable there is NO wide scale deployed voice IMS solution in the market. To my knowledge MANY operators already have wide scale deployed SIP VoIP in the backbone. Sure i give you that op A and B may done this differently. But so what? So long as the inter-operator interfaces exist .. and this can (is) be agreed on a bilateral basis.
The white elephant when it comes to standards is that inspite of the evidence, people somehow expect that things will work automatically and magically because they are adopted. We will have lots of headaches even with the the so called standrads approaches.
In response to "
"Bottom line is the GSMA adoption of onevoice is largely meaningless."
Perhaps, let's wait and see will we? If they had endorsed VoLGA instead would that have been largely meaningless too or would you have seized upon it as another nail in IMS' coffin?
"
I am not interested in knocking nails into coffins. I am still gainfully employed doing other things :))). Neither am i interested in saying VOLGA is the bees knees but VOLGA has its virtues.
The comment that I made has a context, and it is not a context of pitting volga against onevoice. A PR is a piece of paper. A more illustrious paper was waved before adoring masses and the bearer claimed "... peace in our times ...". Millions lost their lives afterwards.
Original anoymous said "IMS is a very large beast precisely because operator networks have complex requirements."
Sorry I disagree. IMS has useful components - i agree. IMS has lots of bits in it that are questionable (in fact in a sense the one voice iniative acknowledges this via pushing for a profile). These may have got through because someone claimed that they were operator requirements. We need to be honest: many requirements are generated by lone individuals with little organisational backing BUT let me ignore this.
IMS had its motivations in particular business models. The market place has changed. But the rational and purpose of IMS has not.
Some professional standards people are trapped in that original thinking and "push" as though the world has not changed. Perhaps it some in the "IMS industry" are also reluctant to acknowledge - in the short term there is still the opportunity to "sell" iron, the buyer of the iron may later wonder whay he has bought ... but the sale has been made and the cash has been received. Livelihoods and futures of companies are at stake ... it is easy to try and imply that there is no alternative to IMS so one must buy it. But the operators are geting savvy. slowly they are seing that parts of IMS may be relevant but they can will do lots without IMS at all ....
Dean i don't diasgaree with the gist of your comment published at 08:04. But i feel uneasy with the statement " ...large operators need very robust and scalable telephony systems. That's why fixed-IMS is making headway for PSTN replacement. ...". i don't think that "fixed-IMS" in a generic sense is making headway. What is actually happening is that component of IMS are being used. Even BT's 21CN in a high profile move put off much of its IMS deployment. I suspect that BT realised that they only need a few bits and pieces i.e. use SIP on the backbone and throw in a few boxes.
A while ago, i read that much of the business case for IMS for a fixed operator came from selling off real estate ... don't know if this is true. But it does bring a smile to one's face ....
The problem is that IMS has been designed without central and upfront consideration of:
- Business models and the ability to innovate new ones
- User experience
- Devices
Conversely, the web comes from exactly the opposite angle, putting user experience and new business models before robustness and overall interoperability.
There are certain aspects of IMS that are valuable - but the problem is that the whole is locked into some legacy viewpoints.
It still astonishes me to hear people refer to "terminals" and "multimedia" as if they've emigrated from the 1980s. I still encounter people who believe that "subscriptions" are the only model of customer engagement.
The problem is that no matter how "smart" the pipe becomes, it's being made to look dumb by the devices at one end, and the cloud at the other. I believe that to be an inevitable consequence of Moore's Law, coupled to IP.
Re: recent anonymous point on fixed-IMS
Fair enough, I'm a mobile-focused analyst so I haven't looked in as much depth at the economics of the fixed side of voice for a while. I'm away that BT is less IMS-centric, but FT and others more so.
My sense has been that "some fixed telephony providers seem to make a case for IMS" - obviously it will depend on legacy infrastructure, overall srategy for services, local competition & regulation etc.
Also, unlike mobile IMS, there aren't the same device & coverage/handover issues, and there's an existing PSTN business to replace (hopefully at lower production/maintenance costs)
Dean
Post a Comment