Informa Telecoms & Media has a series of pre-Mobile World Congress (MWC) interviews up with a roundup of industry luminaries including Virgin Mobile founder Tom Alexander, O2 vice president Mike Short, Bell Labs UK director Louis Samuel, Ericsson CTO for Northwest Europe John Cunliffe, and T-Mobile UK head of technology strategy Tony Weiner, on the theme of "2020 Vision" (heh). The (rather weak) editorial is here.
We won't dwell on the fact that one of the panel predicted that the hit technologies of the future would be 3G video calls and NFC (!), and instead cut to this:
By way of illustration, Bengt Nordstrom [one of the interviewees] takes a punt on the 2020 Mobile World Congress awards. The event, he suggests, will be happening in Beijing. "'Facebook Free Talk & Chat', already the world's most popular voice and messaging service, also wins a GSMA award," he predicts (although it will be surprising if the GSMA hasn't had a rebrand of its own by this stage).
We'd instead go for the following winners at MWC 2020:
And the winner is....IBM Global Services' Multitenant Service Provider Network Outsourcing division. I thank you! An honourable mention goes to BT Openreach.
Why? If you think there's pressure on margins and capacity now, think what it'll be like in 2020. If you think investors are getting snippy about CAPEX now, and pressing you to sign network-sharing agreements, think what it'll be like in 2020. The core-network equivalent of sharing towers, ducts, dark fibre, power, etc, is to move your core network into a massive multi-tenant hosting infrastructure - exactly the sort of thing that IBM, BT Global Services, Amazon, Salesforce, Akamai, Google, the Internet Archive, etc, have built up a core speciality in.
Also, the value of the core network itself is changing. According to Brough Turner, between 96% and 99% of the data traffic on North American mobile networks is heading straight out to the Internet - so if it touches the core network at all, it's got no business being there. Not only will more and more voice and messaging bypass it in future, the price of voice and messaging is in any case marching steadily downwards, and the value of the softswitch must go with it.
Managing really big high-availability data centres and cloud technology is their speciality, just as reliability used to be AT&T Long Lines' profession. If there's no point having four sets of ducts and cell sites, there's not much point having four sets of data centres and four sets of redundant backups.
No doubt there'll be a sexy and overrated Apple product launch at that year's Macworld a couple of weeks earlier, but our pick is the QQ-Skype phone, courtesy of QQ, Skype, and Zhang's Shanzhai & Chips Ltd. It's running Linux, of course, and supports OpenARML-over-XMPP for its augmented reality functions. There is a scramble to get IPv6 deployed in mobile networks after it is realised that its voice capability depends on them all having their own globally routable IPv6 addresses.
We've given our reasons for being keen on QQ before; and despite the general loss of buzz about Skype since about 2006, it's still steadily eating into the international voice market. However, as it's probably unwise to have voice and video traffic do multiple hops over the cellular air interface, the client is probably very much Skype in name only. But perhaps the call control will be in the cloud - Skype's cloud of clients running on millions of fixed machines.
AR is currently waiting only for a widely deployed client and a standard protocol to explode - OpenARML does, in fact, exist, but the clients currently talk proprietary protocols.
And the Shanzhai stuff? Whether it's actually from that community or not, we suspect that the outcome of the great smartphone smashup is going to be very PC-like - a broadly standard architecture that can accept many different vendors' components and many different form factors. The shanzhai devices are currently the closest to that we've got; although there will be some very un-shanzhai materials science stuff in the devices of 2020...not just in screens and batteries, but in things like nanomechanical radio.
Astringent Systems' 59867 Optical Wavelength Router was the undisputed hit of the conference. Kite wind turbines were selling well on the Avenida. Unfortunately, there's no stand for "really huge numbers of bog-standard PC servers".
Down in the packet-pushing world, even now, a lot of people are beginning to find IP and MPLS rather constraining for high-performance applications. National Research and Education Networks (NRENs), as usual, are going first, building backbone networks that consist entirely of private wavelengths directly on the fibre. As we move from virtualising software across racks of servers, to virtualising it across whole data centres, to virtualising it across whole globally-distributed data centre infrastructures, it's increasingly useful to be able to treat the whole system as a single network fabric.
A key barrier to utilising direct lightwave communications is that unless you have diverse fibre routes to all your DCs, you're essentially back in the pre-internetworking world where all your routes are predetermined. The major technical challenge is to switch lightwaves with the same kind of flexibility we have with IP packets - some equipment can now do this, but it has to take the packets through an electronic stage. Just as really high performance IP routers are defined by how much traffic they can route in hardware TCAMs rather than using the CPU and software routines, really high performance optical networking will need to do as much as possible in light rather than having to hit the grey silicon (as opposed to the glassy kind).
Of course, by 2020, power will be everyone's number one issue, whether the climate forces it on us or supply constraints save us from the climate. Available wind power, and the percentage of the time it's available, goes up exponentially with tower height; the well-dressed emerging market base station will probably wear a kite-turbine, and your recovery crews (probably working harder than ever before - the future's going to get weird) will have them as part of their standard load-out. Don't think we're joking- the renewables industry has a surprising number of aerial turbine startups.
All those data centres, of course, will be full of cheap PC-based linux boxes - that's how Google does it now, and we can't expect telcos to catch on in a mere decade.
We reckon, however, that the conference will still be in Barcelona. Nice place, and not too many Mass Group Incidents.
[If you'd like to meet a Telco 2.0 analyst at MWC next week, just contact us]