[LINK] tRe: NBN white-elephant-to-be ...

Paul Brooks pbrooks-link at layer10.com.au
Thu Aug 26 15:53:24 AEST 2010

  On 26/08/2010 3:05 PM, Kim Holburn wrote:
> Here is an article that reckons that we could get 10Gb out of our
> current copper.  Written in 2006 it doesn't seem to have eventuated.
> http://arstechnica.com/hardware/news/2006/10/7952.ars
Note thats 10Gbps aggregate over 50 Cat-3 twisted pairs - not on each one.

I looked at this a lot a couple of years ago, while we were trying to get VDSL2 
permitted in Australia.
Yes, with DSM theoretical models, and even some real-world demonstrations, they could 
get up to 200 Mbps on a single copper pair - a form of VDSL2 on steroids.

Here is a link to a vendor field trial (requires registration): 

and another field trial result: http://www.lightreading.com/document.asp?doc_id=173580

Look up 'vectoring DSM VDSL2' in your favourite search site and you'll find many 
articles about the technique.

In the latter case, they achieved 500 Mbps - at a distance of 500 metres, using 6 
parallel pairs, or a bit over 80 Mbps per pair.

Such speeds are achieved over very short lengths - a few hundred metres at best - and 
depend on having every pair connected to the same line-card on the same DSLAM operated 
by the same company - goodbye ULLS, and competitive DSLAMS.

> Note that the potential bandwidth of fibre is much, much greater than
> this.  I think people have got 69Tb (Terrabits) out of a single fibre
> 240Km long.  I expect that we will be able to get several orders of
> magnitude better than that in the future.
In real world deployments people are getting 10 Gbps per user in WDM-PON deployments 
now, and 100 Gbps per user won't be too far away now the IEEE has standardised those 
physical interfaces.

/straightface - Pretty hard to get a copper pair to do that, let alone to a distance 
of 20 kilometres.

More information about the Link mailing list