Interesting research for Melbourne (long)
Reply: 1.1.1
From: Sim' Hampel
On 12/5/01 12:15:00 PM, Andrew Firmage wrote:
>ADSL will just move the
>bottleneck from the home modem
>to the service providers.
I think you're missing the point slightly... until now there was a real maximum connection speed from homes of 56Kbps due to analogue modems (ignoring other connection types such as ISDN which are way too expensive for widespread domestic use).
With ADSL we remove that last-mile speed limitation (or at least raise the bar significantly for a while).
True, there will be new bottlenecks at the exchanges, but it is relatively easy (I didn't say cheap) to upgrade the capacity at this single point to handle the new capacity requirements.
You can have all the capacity in the world from your exchanges, but if you can't deliver some of that at a decent rate over the last mile (like you can with ADSL) then what's the point ?
The other thing is that you are never expected to get a sustained 1.5Mbps data stream over your ADSL connection. Consider current pricing models with Telstra BigPond for residential use...
Their new 1.5Mbps residential plan has a 5GB download limit per month, and then they charge you 17.5c per additional MB above that.
Let's do some sums...
[Notes:
1. K = Kilo = 1024 (or 1000, if you are Telstra).
2. M = Mega = 1024x1024 (or 1000x1000 if you are Telstra)
3. b = bits
4. B = bytes = 8 bits
3. bps = bits per second
4. Bps = Bytes per second
1.5 Mbps = 1500 Kbps = 187.5 KBps
Now there are 3600 seconds in one hour... so that means we can theoretically transfer 675000 KB per hour, or around 660 MB per hour (wow download a CD each hour !!).
Now by Telstra's definition, 5 GB = 5000 MB, so it would take us only 7.5 hours to use up our 5 GB allocation. So for the rest of the 700 odd hours in the month, we are paying about $115 per hour (at 17.5c per MB) for our downloads. That's a helluva lot of money. Of course this assumes you are sustaining your 1.5M downloads continuously.
The point is that you don't actually sustain those transfers. So your "1.5M * 5000 = 7500Mbit/s of carrier load" is a bit meaningless.
Sure, there will be busy periods and there is indeed a bottleneck at the exchange. And another at the ISP. And another at the international backbone. And so on. But these individual bottlenecks are much easier to upgrade than the old 56K bottleneck at all of those 5000 houses.