Saturday, 2 June 2012

NBN: The Devil in the Technical Details

There's stuff that I don't know about the technical implementation of the NBN using Gigabit PON (Passive Optical Network) - and I'm not sure where to find them:
  • What sort of fibre are they using?
    • Can it be upgraded to normal Ethernet?
  • Does this fibre have a 20km range?
    • Is that 28dB loss or more in 20km?
  • How many dB loss do the splitters introduce?
  • The wikipedia page suggests that the 2.448Gbps downstream link is normally split into 32 subscriber fibres - and that it does that in 28db.
    • Is that a single splitter in-the-field? [low loss, high fan-out]
    • Or a number of splitters that will be deployed between the head-end and subscriber.
  • Will NBN Co be laying additional empty conduit/pipe in the pits it digs?
    • Or if using aerial cables (power poles), making provision to easily run additional fibre.
In 2000, Adrian Blake in Cooma paid AGL/Great Southern Energy to lay extra conduit in the trenches being dug to reticulate natural gas in the town. The "Dot Boom" became the "Dot Bust" before the project could be financed, so we'll never know how that would've worked. Ten years later, Telstra officially acquired the conduit so next-to-nothing as the company assets were sold.

Cooma was near the last town in that AGL project, so no other towns were targetted. Even so, it's unlikely that AGL would've unilaterally laid empty conduit "on spec", ie. without a contract.

Even Telstra, who intimately knew the costs and value of trenching, over the last 20 years didn't mandate empty conduit to be laid in all trenches opened. The engineering staff knew that there was a need in the immediate future for a change from copper to fibre in the customer distribution network, just as they'd made that transition since the first intercity optical fibre laid circa 1986. Yet nobody thought to prepare. How could that be?

The lesson was that nobody, inside or outside AGL and Telstra, had ever thought of the trenching as an asset that could be leveraged... At $25-$45,000/km for trenching [or ~$1M/ in urban settings], that's a remarkable commercial oversight.

So, will NBN Co be laying empty conduit(s) alongside its new fibre?

I think it's hubris to assume that "this network will never fail, be damaged, need expansion/upgrade or need to be changed".
It's an obvious and cheap thing to do, but experience shows that Utilities Engineers and their managers haven't thought this way before.

And why am I dubious of GPON?

Because nobody who does digital networks uses it, everyone uses Ethernet.
That's what all commodity devices use and what chip/hardware vendors invest in and develop for.
We now have 10/40/100Gbps over Fibre defined for Ethernet.

Plus Ethernet-Fibre is available in WDM (Wave Division Multiplexing), allowing the capacity of existing fibres/networks to be upgraded in-place and on-demand. This maximises the utility of the cable-plant investment whilst optimising the investment in capacity with demand. Economically, this is close to a perfect situation.

In the mid-80's, Telecomms Engineers invented ATM (Asynchronous Transfer Mode) and Telco's + their vendors invested massively in building and deploying this technology. It was meant to be the future technology to underpin all converged digital services.
But like Fibre Channel, it's a technology that's obsolete and on the wane.

Backbone networks are universally being migrated away from ATM to IP and Ethernet. Whilst IP/Ethernet don't yet have all the QoS and manageability features large Telcos like, they are appearing.

How long before GPON is similarly out-moded and in dire need of upgrade? As soon as 10 years?

Will the NBN Co network allow that upgrade cheaply and easily, or in another 10 years will we have to spend another $20-$40B digging exactly the same trenches?

As said at the start, I don't know where to find out this technical and design information.
Any pointers welcome!

No comments:

Post a Comment

Note: only a member of this blog may post a comment.