Thursday, 18 April 2013

NBN: DSL's Four Horsemen of the Apocalypse

There are some major project variables for a DSL NBN.

Let me predicate this piece with a statement of my position:
I'm not in favour of either an FTTN or FTTP NBN, but of a reasonable National Digital Infrastructure. Australia has needed an NBN since before Trujillo floated the idea in 2005, but for complex Politico-Economic reasons, it hasn't happened. Ideally, after the deregulation of Telecommunications and the privatisation of Telstra, the private sector would've co-operated and delivered Universal Fast Broadband. It didn't. Nothing focuses the mind of Engineers so well as a strict Budget and economic necessity: while a Publicly funded network can be built, we know it may not be the most cost-efficient engineering design because "Necessity is the Mother of Invention". I'm sure both side of politics would rather be devoting their energies and $30 billion of public money to other priorities, not making good on a gap that the private sector should've filled.
I'm putting aside the cost of upgrading an FTTN to Fibre-to-Premises. That's another debate.
Also, I've previously identified that Telstra poses a "drop-dead" risk for an FTTN: the shareholders could easily decide to charge more to sell or lease the "last-mile" copper.

Turnbull in his plan has omitted two critical factors:
  • An estimate of the total cost of the DSL network,
    • $900/line is tucked away in there, we can infer from "25% of Fibre Costs" and "Mac Bank estimates $3,600/premise for Fibre".
  • The Node-Premise maximum distance. Is it 1500m, 800m, 400m or "varies between 1500m and 400m"?
There is a huge amount of good work out there on costing, impact and Economic Benefit of an FTTN, but only one source on an FTTP: NBN Co. Their figures haven't been tested and until delivered, we don't know how good their estimating has been.

There are subtleties in the financing and economics of big projects that mean even if the CapEx of an FTTN is the same ($29.5 billion) as FTTP, it is still a better investment: the time value of money, expressed in the Net Present Value (NPV).

There is also the Engineering Risk and estimation error inherent in the FTTP figures We just don't know what the real project cost will be. The high-volume construction hasn't started, we don't know if their per-premise cost is reasonable or not (I make their figures to be $1.7 billion common infrastructure plus incremental $2195/premise).

All large projects contain "surprises": until you've done, you can't tell what you don't know.
My experience is that almost all "surprises" are bad: they cost more and delay things. Sometimes you cut a break, but not often.

It's not unreasonable to believe that NBN Co will over-run on budget for the Fibre rollout. The only question is: by how much? We'll only know in 2014 after their first full year of full-rate rollout.

Schedule I don't have a problem with... Fibre is laid by independent teams and can scale up linearly: teams don't get in each others way nor depend on others work, once the commons are laid down.

It may be cynical, but I think ex-Servicemen from the USA might be a great source of expert technical folk for the fibre jointing. With the war winding down, they'll be a lot of them getting out, with very high-level skills and no work of that sort in the USA.

The Four Horsemen for the DSL-NBN:
  • Copper Remediation
    • The 2005 Telstra NBN for 4M of 5.6M lines in the Big 5 cities, 20,000 nodes for 2/3 lines, included $3.1 billion from Telstra and a preferred position of 6Mbps at $5.7 billion total.
    • Trujillo gave Howard in 2005 (TLS385) 3 options for Govt contribution:
      • $1.1 billion for 1.5Mbps everywhere
      • $2.6 billion for 6 Mbps (ADSL2)
      • $4.7 billion for 12 Mbps. [The Rudd DSL-NBN figure]
    • They also rated "remediation" as one of the big project costs.
    • Since 2005, the network will have deteriorated further. But by how much?
      • Will this be another $5-$10 billion on top?
  • NTD's (Network Termination Devices)
    • These are specific devices to NBN Co bitstream, not your usual ADSL router/modems.
    • They are needed on all network types, Wireless, Satellite and Fibre.
    • I've not seen costings for them or cost to install.
      • My guess for Fibre is $250/unit in volume
      • And VDSL, $200/unit.
    • I don't know if NBN Co will be charging an install fee to defray their expense.
      • I think NBN Co cost will be $250 device + $150 labour: $
    • Every DSL NBN plan I've seen, including the Coalitions', ignore the cost of the supply and install of the NTD. I infer from this that it's a customer expense, paid for at connection time.
      • because it's part of the Telco network, the NTD has to be installed by a licensed cabler.
      • My guess would be $200 device + $250 labour (retail rate, 1hr) = $450 install.
  • Fibre uplinks and SFP GBIC's
    • My rough estimate was for 75,000km of Fibre to node. Access Economics in 2009 estimated 81,000 km.
    • My rough estimate for laying single tube fibre in ducts was $20,000/km. An expert opinion is $17,500 to $25,000. Splicing is time consuming and expensive. Need a person at both ends to check splices are good.
    • I'd forgotten that high-power, Extended Distance Ethernet Fibre interfaces (SFP's or GBIC's) are needed. Two per node for redundant links over diverse paths.
      • I think longest fibre run to a node will be ~25km, though with regeneration, this may be reduced to 10km.
      • High-power GBIC's are still expensive.
      • 1Gbps can be used to start, but will
      • 10Gbps or 40Gbps are needed to get beyond 10Mbps per line with 100 ports in a node. They're currently $1,000's, even in volume, for LX and EX versions.
  • Relaying and rejoint the copper network for an optimal Node layout
    • Telstra in responding to the FANOC proposal said they could use 40% fewer nodes by redeploying the last-mile from "best for phone" to "best for DSL".
    • This comes at a price

No comments:

Post a Comment

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