[LINK] terrabit per second internet

Richard Chirgwin rchirgwin at ozemail.com.au
Wed Sep 9 11:41:19 AEST 2009

Paul Brooks wrote:
> Roger Clarke wrote:
>> At 21:51 +1000 8/9/09, David Boxall wrote:
>>> What impressed me was to hear Aussies pronouncing router as 'rowter'.
>>> When I was studying (just this side of the Jurassic) it was 'rooter'.
>> As one of the rapidly declining band of guardians of Orstralian as 
>> she wuz spoke ...
>> ...  the word 'route' has been pronounced 'rowte' in the Australian 
>> military for a very long time.
>> It's very difficult to sustain credibility in front of your troops if 
>> you say something like 'and for your final exercise prior to 
>> graduating as a driver, instead of me planning where you're driving 
>> tomorrow, you can all route yourselves'.
> Yet oddly, in the military term 'route march' is always pronouced 'root 
> march', while a standalone 'route' prononced 'rowte' is a hasty retreat 
> - and when an army is 'routed (rowted)' and runs away they are probably 
> 'rooted' as well.
> Also, while router = rooter is through to be Australian, the router that 
> makes moulded edgings in timber has always been pronounced 'rowter' in 
> our house (from well before packet-directing routers were invented).
> Anyhoo, back on topic (terabit-per-second optical switching, if you've 
> forgotten the subject line by now!) for a minute...
> I ran a few calculations recently for a conference talk on 100Gbps 
> optical transmission technologies...
> Did you know that at the speed of light in glass, a (todays) standard 
> 10Gbps optical signal switches the laser signal on and off  54 times 
> before the first pulse has travelled just 1 metre through the fibre?
> A  40Gbps optical signals (220 pulses per light-metre, or just 22 
> picoseconds between pulses) appears to be approaching a limit for 
> practical long-distance communications - the emerging 100Gbps 
> technologies are falling back to 10 parallel channels of 10Gbps, or 
> smarter use of 25 Gbps signals, but not any sign of real 100Gbps optical 
> signals being viable, because dispersion in the glass smears the pulses 
> together before it has travelled more than a few kilometres down the fibre.
> Yes, the magic crystals may be able to switch terrabit per second 
> signals - but those signals will only be able to travel a very short 
> distance before they become smeared together, and on long-haul fibre I 
> suspect the bitstream will need to be split across many much slower 
> bit-rate channels to have much impact in the Internet.
You're right; the terabit optical switches are more aimed at Internet 
exchange systems.

As you say, you'll need ten fibres (for example) to run a terabit Sy-Me; 
but the passives that CUDOS is working on (not only them, there are 
plenty of other researchers in this particular game) would, for eg, 
split / combine the terabit "trunk" across the multiple bits of glass 
without electrical conversion.

In talking to Prof. Eggleton last year for ASOT (self promoting link: 
<http://itradio.com.au/networking/wp-content/uploads/ASOT57.mp3>) I was 
also struck with the power-saving angle...
> (Telco Trivia: did you know that a 10Gbps optical channel between Sydney 
> and San Francisco acts as an optical delay-line memory, storing 745 
> Mbytes of data as photons inside the glass before the first bit pops out 
> at the far end?)
Wow ... I was impressed back when an IBM-er told me there was about 10 M 
of data in the delay-line between Sydney and Perth!

> Cheers,
>     Paul.
> _______________________________________________
> Link mailing list
> Link at mailman.anu.edu.au
> http://mailman.anu.edu.au/mailman/listinfo/link

More information about the Link mailing list