Senator Stephen Conroy has just released the NBN Co Business Case Summary [3MB PDF]. I’m about to start reading it, and may or may not have some comments later.
7 Replies to “NBN Co Business Case Summary”
Comments are closed.
Word-whore. I write 'em. I talk 'em. Information, politics, media, and the cybers. I drink. I use bad words. All publication is a political act. All communication is propaganda. All art is pornography. All business is personal. All hail Eris! Vive les poissons rouges sauvages!
I don’t like the conservative projection in the demand for speeds. They’re ignoring the potential speed demands from new applications. Option A2 (capped at 100Mbps for eternity) is clearly a strawman.
What about the growing demand for upstream speed?
The document includes some slides copied from a PowerPoint. Slides are numbered 2, 3, 5, 6, and 7. I wanna know what is on the missing slide #4?
From memory the planned NBN architecture does not have the capacity to deliver ubiquitous committed downstream speeds of 1Gbps due to the hardware at Fibre Access Nodes etc.
I assume upstream is similarly restricted.
If everyone on a local Fibre Access Node wants a reliable committed speed above 100Mbps then further investment would be required.
I’m flat out today and tomorrow, so I may not be able to response to comments quickly or in detail. But for now…
I asked NBN Co to clarify this paragraph from page 18:
An NBN Co spokesperson replied:
So with that background…
@Rick: You’re right. A GPON network splits a fibre amongst a number of users. NBN Co’s is being set up so that everyone will be able to be given 100Mbps as a committed speed — which will translate into sustained speeds for everyone of something like 70Mbps, from memory. Individual customers can be set up with 1Gbps peak speeds — but presumably not all.
That said, I daresay it’d be possible for NBN Co to run a dedicated fibre to any customer who needs it, just as Telstra can do the same now. There’d presumably be a difference between what NBN Co has to offer to all premises as a base product — what they’re funded to install as part of the national roll-out — as opposed to what they can choose to offer as an up-sell.
@Greg: Those slides are familiar. I’ve seen them in the slide deck that Mike Quigley shuffles for different presentations. I reckon if you look in the relevant bit of the NBN Co website you’ll find those presentations as I know they’ve been posted. I just dont have time to chase them right now.
Copying over my “back of napkin” maths from http://delimiter.com.au/2010/10/22/nbn-war-is-the-australian-out-of-line/#comment-27012
Assumptions:
– NBN Co will deploy GPON
– each fibre upstream of the splitter will support 32 ONTs
– all 32 customers have subscribed to the maximum speed plan available
If all 32 customers are utilising as much bandwidth as they can have then:
– downstream – 2.4 Gb/s / 32 ~= 75 Mb/s
– upstream – 1.2 Gb/s / 32 ~= 37 Mb/s
These speeds would be the maximum committed information rate (CIR) that NBN Co could reasonably wholesale. Any speeds above that would include a component of excess information rate (EIR).
However, reality is likely to a bit different than that.
Not everyone will sign up for higher speeds, they might just appreciate the reliability and low latency.
Not everyone will use as much bandwidth as possible all the time. Effective information rate will be somewhere between the CIR and their plan speed depending on the simultaneous utilisation by others. If there is only 1 customer online and their plan allows it theoretically they could use the full bandwidth provided by GPON.
I expect that will meet our needs for about 20 years. We could then replace GPON with whatever version of PON is mature and economical at the time. Given that a large proportion of the cost of the NBN is in rolling out the fibre, upgrades at the edges will offer a cheaper and easier increase in performance.
“Not everyone will use as much bandwidth as possible all the time.”
That’s _probably_ true, but may be ignoring new uses we might invent when higher bandwidth becomes available.
Some back-of-an-envelope calculations I did earlier today suggests that the entire output of the recorded music industry is only around 2.5 terabytes a year. That’s _well_ within reach of people having that plugged into their living room media player – it’s only a couple of hundred bucks of hard drive – plus 60 or 70 gig a day of bittorrenting it down, presumably with the balanced 60 or 70 gig of bittorrent seed uploads too…
And it’ll only be 5 years until Officeworks are selling drives big enough for regular people to think “$200 for a 100TB external drive? I could grab one of them and store every new movie released in the last year!”
@Martin Barry: Thank you for that. It all seems spot on. That the NBN will use GPON has been confirmed in their technical presentations. The rest is just arithmetic and matched the conversations that NBN Co is having.
TCP/IP is not the only method available, but it is easily the most common. You cannot use the higher bandwidths in a single down load because of the design of the TCP/IP stack. You can get stacks that transmit more data, but if you are pulling data form a remote location and using a standard stack you may be limited to as little 1 MBit/s. This is distance dependent (welll really delay dependent). Your really would start to notice, that all other things being equal, sites in Australia allow higher bandwidths than sites overseas. You can get over this by running multiple down load sessions. Though this does not help if you have one big file you want to move. I suppose what I am saying is that it is actually quite hard to use such large bandwiths for any lenght of time.