One of the most frustrating aspects to the debate about future directions for Australia’s broadband infrastructure is the complete obsession with headline download and upload speeds that many people will never use at economic price points, to the detriment of any discussion about other key aspects of the Internet experience such as latency, which affects pretty much every Internet user, no matter how modest their needs.
Roundtrip latency—exacerbated by the increasingly inefficient dimensions of TCP communications and the physical distances those communications traverse—is an important determinant of the quality of one’s Internet experience. This is especially the case for real-time applications such as voice, gaming and financial trading.
Given the economic value of comms to the financial trading sector, it is unsurprisingly at the vanguard of doing something about latency. Fibre optic connections between the two American financial centres of Chicago and New York City typically operate at latencies of between 13 and 15 milliseconds. This isn’t good enough for some traders.
Contrary to popular opinion, fibre optic cables actually slow down communications to about two thirds of the actual speed of light in a vacuum such as space due to imperfections in the glass. By contrast, wireless signals travel much closer to the genuine speed of light. Hence two microwave networks—one operated by Epsilon Networks, the other by McKay Brothers—are serving this need by deploying links between Illinois and NY that operate at latencies of as low as 8 milliseconds. 5 milliseconds might not seem like much of a difference but the graph here illustrates the economic value of timeliness in trading.
Of course, wireless communications suffer from weather and attenuation issues not faced by fibre. But as pointed out in Wired magazine, “If your link is only 99 percent reliable, you don’t make money 1 percent of the time; if it’s slower than the competition, you don’t make money 100 percent of the time.”
There are many positive developments occurring in Australia that will serve to improve the local Internet experience. Probably the most significant is the construction of multiple data centres, diffused across population bases such as Perth, Melbourne and Brisbane in addition to the traditional international data offload and upload points in Sydney.
Yesterday’s announcement by Rackspace that it has launched an Australian data centre is another big step in the right direction. Even the controversial decision for the NBN to deploy 121 points of interconnect instead of the preferred 14 has its end user benefits—it again potentially shortens the physical distance, and, thus, latency of Internet communications outside of capital cities.
More can be done.
When one looks around the international comms scene there are two significant omissions in Australia’s infrastructure—the poor use of the waters around and the skies above our sizable island nation to facilitate better communications.
The sea distances between Perth, Adelaide and Melbourne & Brisbane and Cairns respectively are smaller than the road distances—and, thus, the implied routes of extant terrestrial trunks with the potentially superior latencies of undersea alternatives.
Many other countries with large seaboards use underwater festoon cables to connect their coastal cities. Notable examples include Italy, China, the east coast of Canada and Venezuela. Seas have a nice advantage over terrestrial rights-of-way: they don’t charge rent and are cheaper to deploy in.
A new southern cable branching to the population points of Western Australia, South Australia and Victoria could also form a key middle leg of a greater Asia-Australia-Americas route which would provide important redundancy for global Internet users. Both the redundancy and latency advantages could prove key selling points in mounting a business case for such a cable.
The other omission? There is no reason why a subsidised, welfarist program akin to that of the NBN but less costly shouldn’t make use of barely used hundreds upon thousands of class licensed or cheap megahertz to provide high speed point to point wireless broadband on an ad hoc needs basis to rural and remote Australians with genuinely high-end bandwidth requirements. Surely this would be superior to the market crap-shoot that is NBN Co’s limited speed LTE deployment.
Contrary to the demonisation of wireless by some, which posits giant towers in every cul-de-sac compelled to carry the nation’s entire Internet traffic down one 20MHz carrier, there is actually an enormous abundance of spectrum that is fit for purpose for sporadic islands of high-end demand. It is notable that one operator which leverages this opportunity, BigAir, is one of the fastest growing in the sector.
But I am a mere commentator: there are many entrepreneurs and innovators who no doubt have identified other superior ideas that could improve the quality of the Australian internet with minimal expense and risk.
MORE NUANCE? It’s time we had a more nuanced discussion about Australia’s broadband infrastructure. Often when I am in conversation about this topic with members of Australia’s telecom chattering classes, the discussion is always ultimately hijacked by a request that I affirm my support for the all-fibre, all-sparkling new NBN, as is, without qualification and as a marker of my digital savvy.
Such discussions are ultimately a pointless exercise: the NBN debate has become little more than a public test of religious affirmation about the desirability of a specific last mile access medium and the associated conflation that one must support it or be an enemy of the future.
This belies the reality that the NBN, as proposed now by Messrs. Quigley and Conroy, is a highly heterogeneous network which employs at least five distinct access media to my eyes: FTTH GPON, FTTH P2P, LTE, an interim leased lower speed satellite service and a long term higher speed satellite service.
A change of government and we could add a sixth type to that equation: FTTN.
Add to this the fact that there are now twice as many wireless subscriptions to the Internet over GSM, WCDMA, HSPA+, LTE and other untethered technologies as there are for fixed comms, and the Australian Internet and PSTN is an increasingly heterogeneous beast, not less of one. Try finding acknowledgement of that in what passes for “broadband expert opinion” in this country.
Witness this week’s debate about the allegedly high maintenance costs of FTTN compared to FTTH, sparked by BIS Schrapnel and fanned by business commentator Alan Kohler. Under the current NBN plan, the most expensive part of the copper network—that in rural and remote Australia—will be retained and funded by industry levies amounting to nearly $300m annually—nearly half the alleged cost of maintaining the entire national copper network today.
And of course, under FTTN, the most fault prone parts of the copper network—the bundles of copper that feed into exchanges, not individual access lines—would be replaced by fibre.
Given the big cost unknown of the NBN is the state of the Telstra ducts into which it wishes to deploy fibre anyway—some of which are potentially unusable without costly remediation— the biggest omission in this debate is hard data about the actual costs of real plans that appear to be guaranteed funding with no recourse to budget limits or opportunity cost.
You wouldn’t know it from the confident assertions and purported “fact checking” obsessions of some, mind you. For example, one looked long and hard this week for much reporting of the Motion Picture Expert Group’s breakthrough in compression technology which will reduce the bandwidth requirements for video transmission in half.
This presumably has some relevance to a debate about a $40 billion national infrastructure designed primarily to distribute video but you wouldn’t know it reading what passes for broadband expert opinion in this country.