Monthly Archives: September 2012
Communications minister Stephen Conroy has threatened to build a new submarine cable to the United States if the market does not reduce prices from current levels.
His comments came as he rejected European proposals in front of the International Telecommunications Union to introduce a new sender party pays regime for international internet settlements, stating that consumers pay too much for the Internet already.
Speaking at a Columbia Institute for Tele-Information conference in New York City last Monday, Conroy told attendees that “if the international market doesn’t improve [prices], for $250m out of a $40b [NBN] budget I’ll build a link to the US to bring prices down.”
He added, “there are a couple of commercial prospects [on that route] so that’s why I don’t want to leap in. But we pay too much and that’s why I do not support the proposal to charge (for Internet settlements) between countries.”
It’s not the first time that Conroy has flagged concerns about international bandwidth prices, but industry sources were surprised he still thinks it is an issue after several years of steep price cuts.
One telco CEO, who asked to be off the record, said that a greater concern was the NBN’s own proposed connectivity virtual circuit price of $20 per Mb, which would be a larger factor in end user costs than international capacity. Australia-US capacity costs have fallen to around $25 per Mb typically, with one recent deal rumoured to be priced as low as $8.
Conroy’s speech was his most candid yet in explaining the reasoning behind his NBN policy, with him acknowledging that it was “controversial” and “radical.”
He said the NBN model could not easily be replicated in other countries because of Australia’s unique system of federal power over telecommunications.
UNFETTERED POWER: “The regulation of telecommunications powers in Australia is exclusively federal. That means I am in charge of spectrum auctions, and if I say to everyone in this room ‘if you want to bid in our spectrum auction you’d better wear red underpants on your head’, I’ve got some news for you. You’ll be wearing them on your head,” said the minister. “I have unfettered legal power.”
But Conroy emphasised that it wasn’t because of the power he had that he was building an NBN but because of the “responsibility.”
He said that only a government could take the long view for such infrastructure projects, drawing an analogy for his audience with the US federal highway system.
Another motivation for the government-led build was the “bleak view” for telco financial prospects, saying that the “profits just weren’t there anymore.” He attributed the failure of the NBN Mark 1 FTTN tender in 2008 to the fact it closed just a month after the collapse of Lehman Brothers and that none of the bids offered value for money.
Yet another factor behind the NBN build was the actions of the incumbent Telstra, who he said still made 92% of all profits in telecommunications some 15 years after liberalisation. “Telstra had offered an upgrade to FTTN for about 50% of homes but they would have charged $150 a month for 24 Mbit/s,” he said, an allusion to its 2005-era proposals in the last term of the former Coalition government. He also accused Telstra of earlier driving its “challenger”—presumably Optus Vision—bankrupt in pursuing a parallel HFC build in the 1990s, which led to an outcome where only 28% of households in Australia were passed by cable.
Conroy acknowledged that Europe “are not big fans of my model… I’ve been offered money not to come and talk about the NBN there!”
“We are only requiring the NBN to make 7% [return] a year and no self-respecting telco will get out of bed for less than 20%.” Conroy emphasised that the NBN had already signed 40 retail service providers and that end retail prices varied from between $40 and $100 per month for services between 12Mbit/s and 100Mbit/s.
Answering a question about the fixed versus wireless debate, he also said “if you want to think of the NBN as the biggest piece of backhaul for a wireless network then go for it.”
TURNBULL RESPONDS: Shadow communications minister Malcolm Turnbull reacted to Conroy’s speech last night stating “As Mark Twain said confessions are good for the soul but bad for the reputation but never more so than when a cabinet minister confesses to rampant megalomania.”
“He has confirmed in these remarks what so many have said about him; that he is a control freak who despises the private sector industry and regardless of the cost to the taxpayer is determined to use every element of his legal powers to increase the power of the government at the expense of the private sector.”
COMM BANK TO BECOME A TELCO? In an interesting aside, Conroy also hinted that the Commonwealth Bank was considering bypassing telcos and becoming a telco in its own right to go direct to customers. He also suggested that Australia supported the current system of Internet governance and that no case had been made as to why that should change—statements that provide support for the US as it resists attempts from European, Asian and Middle Eastern industry and government interests to radically overhaul international telecommunications regulations and governance.
And while Geddes – an alumnus of industry heavyweights like BT, Sprint and Oracle – casts the shift towards a quality-centric approach to networks as “mathematically inevitable” for operators, he also says that the onus will be on regulators to understand the coming change and moderate behaviours accordingly. Geddes has previously argued that, as the explosive proliferation of diverse devices and apps bombards networks with unprecedented numbers of concurrent and increasingly less isolated flows, a growing variability in loss and delay is reducing the ability of TCP/IP to cope – eventually overwhelming those networks, even if more raw bandwidth is added in an effort to compensate. He posits a basic issue with the nature of supply and demand on broadband networks.
“Historically, the way the telecom industry has worked is it went out and built supply, as with the PSTN, at a fixed quality, and then it set the marketing department free and said ‘go out and hunt for demand’,” Geddes told CommsDay. “But on the internet, the moment you start finding demand, it starts to destroy the properties of the supply; the quality of delivery goes down monotonically with more customers. And you get to a point where the quality degrades enough that you feel obliged to invest in more capacity. But the problem is that the cycle time of that investment drops… a necessary by-product of you multiplexing stuff harder is that the return on reducing serialisation time in a system where the contention is the bottleneck becomes less and less.”
That, of course, means that the return on investment for capex to boost bandwidth – everything from 3G and 4G mobile infrastructure upgrades, to spectrum auctions, to national broadband network projects – starts to drop off. To solve the problem, Geddes is urging operators to drastically rethink their business models. “What you have to manage in the network is not bandwidth, but contention,” he said. “And you flip the whole problem on its head by saying ‘rather than how do I allocate work to queues, how do I allocate the loss and delay that comes with contention?’ And when you think about the network that way, you can say ‘ok – in which network elements, or where along the path, is the contention actually occurring?’ You isolate where the problem is… then manage the contention at that point using existing quality mechanisms or new, clever maths to do it better. I’ve been working with some mathematicians in the UK who’ve been using this stuff in Tier 1 fixed and mobile operators – to somewhat spectacular results.”
While that might sound superficially like existing quality of service management, Geddes explained that it’s actually a fundamentally new set of algorithms. “QoS is allocating priority, this is about allocating the ‘holes’ – how do I trade loss and delay between flows,” he said, “You can take any sub-sequence, any path along the network and pre-contend the traffic, so it doesn’t self-contend downstream.” “And what it takes you towards, ultimately, is from a monoservice to a polyservice network. Today, when people talk about multi-service networks, they’re really multiple monoservice networks; you get voice, video and data, but all three of those come with a fixed quality. A polyservice network is able to allocate different statistical bounds on loss and delay to every flow, and trade across between all of them. That’s the difference – and it’s like the difference between black and white or sepia, and colour.”
“Monoservice networks are unsustainable; polyservice networks are a mathematical inevitability. You can’t get quality and abundance by going back to a circuit model (à la IMS); you need to work directly with the fundamental properties of multiplexing to build contention-managed networks.”
Geddes acknowledges that such a sea-change will present a host of challenges to operators – even if ultimately necessary for their survival – such as rebuilding circuit-centric OSS/BSS systems, and pricing across multiple service levels. But he also sees a key role for the regulator in the paradigm shift.
“We’ve been talking to Ofcom, for example, in the UK, about the need to change the basis on which this market is measured,” he said. “ So just like the car markets went from… emphasizing speed, and then it became about fuel economy… in telecoms, it’s gone speed, bandwidth, bandwidth, speed, [but] the next phase is one of ‘fitness for purpose’, which is kind of like fuel economy.
“Because you can have two networks that offer the same bandwidth, one of which is totally useless, and one of which is wonderful – because of different contention effects. So I think the regulators need to educate themselves on the changing nature of what constitutes fitness for purpose.”