Towards high-capacity fibre-optic communications at the speed of light in vacuum

research being done to use photonic-bandgap rims to make fibers that would allow propagation approaching vacuum speed [2], and if deployed eventually that could drop latency by another 40-60ms. But at least back then there seemed to be enough loss (3.5 dB/km) that it'd be impractical over long distances without excessive boosters, even ignoring any cost/durability issues of the fiber itself. That may have been improved in the last 4 years though, but I'm not sure if it ever did make it to market (or if there are even plans still in progress for it, raw throughput was impressive too FWIW). At any rate: the NZ ISPs seem to be offering pretty solid service on that front even internationally, at least in terms of pure latency. If there is increasing congestion or other issues those are just matters of capex, and between the ISP's servers and premise links there might be more lag being added in. But I don't see any glaring core inefficiencies there in terms of the international link part, unless I'm missing something entirely. Looks a lot better then I'd have expected taking a guess based on how it was a decade ago. ---- 1: https://truenet.nz/international-latency 2: "Towards high-capacity fibre-optic communications at the speed of light in vacuum": https://www.nature.com/articles/nphoton.2013.45 (DOI: 10.1038/nphoton.2013.45)

'vectoring', more active processing of communications in transit (to calculate the correct cross-talk cancelling noise to push down all lines). In addition, wave-division multiplexing with latency at 99.7 per cent the speed of light in a vacuum has been achieved on fibre (admittedly in a lab/academic setting) over a distance of 310 metres (source: http://dx.doi.org/10.1038/nphoton.2013.45). Why is lower latency desirable, from a national economic perspective? Because it creates larger scope for distributed computation of non-parallel processing and higher IO for geographically distributed storage. So data-centres can be located on cheaper land, further away from the CBD. Perhaps in future it also means someone in Darwin could utilise the idling processor, or spare hard disk space, of someone in Sydney via something like Etherium. All of this becomes harder and less efficienct as latency creeps up. For general background, see: https://en.wikipedia.org/wiki/Fiber-optic_communication#Comp... On future upgradability, I'd argue we're reaching the upper bandwidth limits of what copper can provide (as a real world, fixed line infrastructure). This seems to be around 100-150mbit non-symmetric under the most ideal conditions, using some pretty fancy tech (on both the carrier and consumer ends), although the hypothetical maximum is supposedly around 10Gbits (achievable in LANs, unlikely to be achievable in WANs with current copper assets). How long until this is not 'enough'? In contrast, for single strand fibre, NEC have achieved 101Tbit/s over 165km, and NTT 69.1 Tbit/s over 240km. Cisco projects the volume of global IP traffic to grow at a compounding annualised growth rate of 23% between 2014 and 2019; in other words global IP traffic volume will triple in 5 years (source: http://www.cisco.com/c/en/us/solutions/collateral/service-pr...). Keep in mind this is strictly global IP traffic. Alcatel-lucent and Bell Labs, factoring in 'metro' (non-backbone) traffic (e.g. office to datacentre, consumer to CDN), forecast a quadrupling of bandwidth volume between 2015 and 2020, with an apparently exponential growth rate (source: https://mc-11560-1882305812.us-east-1.elb.amazonaws.com/site...). By opting to 'destroy' the original NBN, our government has knee-capped Australia's future economic prospects. Perhaps Lee Kuan Yew was right and Australia is destined to be the 'poor white-trash of Asia'.