I was reading about Internet routers a few months ago and "the world's
leading expert in Internet routers" said that the speed of Internet routers
had followed Moore's Law religiously since the mid 1980s.

I also read that the theoretical bandwidth of a single optic fibre is 10
Tbps, so I'd imagine it's safe to say that fibre can handle anything you
throw at it.

So has the cost of Internet bandwidth fallen in line with Moore's Law since
the mid 1980s? And is it a reasonable assumption that if router speeds
increase by a factor of X, Internet bandwidth becomes X times cheaper?


--
Steve - www.digitalradiotech.co.uk - Digital Radio News & Info

The adoption of DAB was the most incompetent technical
decision ever made in the history of UK broadcasting:
http://www.digitalradiotech.co.uk/da...ion_of_dab.htm

DAB sounds worse than FM wrote:

Nope. I haven't noticed any comms phone bill reduced to a 1000th of the
cost in 1985. Perhaps people are doing different things with the
bandwidth - new applications are possible (video mainly) that were not
possible before, but we are not saving great amounts of money as such as
consumers. And it's probably true for the owners of the routers running
the backbone of the net.

And is it a reasonable assumption that if router speeds
Nope, a standing charge will be applicable regardless of the speed and
quality of service. There will however be pressure to reduce the
standing charges but it won't track with whatever cost benefits the
technology is reducing by. There is a limit.

For instance, look at memory card manufacturers. The price of flash has
fallen so much that they are now having to cut back on previous fixed
packaging costs - cardboard and printed leaflets are small in cost to
produce, but as a percentage of the overall cost of the finished product
when taking transport in consideration - that's rising and looking
embarrassing. Get ready to dust one out your cornflakes packet any day
now....


--
Adrian C

In message <63j2lnF27ibqgU1@mid.individual.net>, Adrian C
<email@here.invalid> wrote
1990 approx. 5p/minute at 28,800 bps

2008 approx £15 for 43,000 minutes at 6,000,000 bps

The cost per bit per second has reduced by around x30,000 in the past 18
years.



--
Alan
news2006 {at} amac {dot} f2s {dot} com

Alan wrote:

That told him! :-)


--
Steve - www.digitalradiotech.co.uk - Digital Radio News & Info

The adoption of DAB was the most incompetent technical
decision ever made in the history of UK broadcasting:
http://www.digitalradiotech.co.uk/da...ion_of_dab.htm

DAB sounds worse than FM wrote:

Er... No.

:-)

--
Adrian C

Adrian C wrote:

I thought it would have been obvious from what I wrote that I wasn't talking
about consumer speeds.



But is it true for the owners of the routers running the backbone of the
net? That's the question. For example, if they installed a router ever 2
years and every time they installed a new one the bandwidth had doubled,
that would suggest that Internet bandwidth cost should fall in line with
Moore's Law.



The cost of the Internet router itself would be equivalent to the "standing
charge" you're talking about, but that fixed cost is depreciated over its
lifetime, so you can include it in an overall average cost of bandwidth.


--
Steve - www.digitalradiotech.co.uk - Digital Radio News & Info

The adoption of DAB was the most incompetent technical
decision ever made in the history of UK broadcasting:
http://www.digitalradiotech.co.uk/da...ion_of_dab.htm

Adrian C wrote:

You need to try Skype then! :-)


--
Steve - www.digitalradiotech.co.uk - Digital Radio News & Info

The adoption of DAB was the most incompetent technical
decision ever made in the history of UK broadcasting:
http://www.digitalradiotech.co.uk/da...ion_of_dab.htm

On 9-Mar-2008, "DAB sounds worse than FM" <dab.is@dead> wrote:

The OP quoted the speed of routers had followed Moore's Law, that
isn't the same as saying the price of routers has fallen religiously
in accordance with Moore's Law. They are expensive to replace,
and don't get upgraded on a yearly cycle.
Again it's not about a single optical fibre, there is plenty of dark
fibre that could be used for extra bandwidth, rather than try to
push 10Tbs over a single fibre.
The other issue is does the cost of supplying the service
(ie bandwidth) reflect the true cost of providing it including
premises, staff, electricity, lighting, heating, sales, billing,
, the cost of providing LINX which many ISPS's
connect through, etc?
Or maybe it's like many other industries a case of what
the public is prepared to pay.

DAB sounds worse than FM wrote:
at the terminal was the basis of charges for information independant of
the number actual data bits that made up that information.

Well, it seems you are correct. Some one has beaten you to name it.

<http://en.wikipedia.org/wiki/Dark_fiber#Butters.27_Law_of_Photonics>

:-)

--
Adrian C

Adrian C wrote:

Sorry. Brain's out to lunch.... You started this topic with a reference
to probably the same bloke...

--
Adrian C

ato_zee@hotmail.com wrote:

Sure.



My assumption has been that the replacement of the routers would be included
as a fixed cost (the cost of the router depreciated over time) in with all
the other ones you've listed above. So if the cost per unit bandwidth is
given by:

cost per Mbps = total fixed costs including router replacements / total
bandwidth

So if the total fixed costs including replacing routers was constant, but
due to Moore's Law increasing the routers' bandwidth the total bandwidth
served went up in line with Moore's Law, that would mean that the cost per
unit bandwidth would go down in line with Moore's Law as well.



I'm thinking more about the bandwidth costs for heavy users, such as the
BBC.


--
Steve - www.digitalradiotech.co.uk - Digital Radio News & Info

The adoption of DAB was the most incompetent technical
decision ever made in the history of UK broadcasting:
http://www.digitalradiotech.co.uk/da...ion_of_dab.htm

Adrian C wrote:

No, the bloke I was thinking of was Nick McKeown who received a prize from
the British Computing Society, and he was described (IIRC) as the world's
leading expert on routers, so I dug out a couple of things he'd written,
such as this:

http://pnrl.stanford.edu/light_flow/BCSv6.0.ppt

(page 49 has got a graph of the log of router speed vs time)

I think that Butters Law is about how much data it has been possible to push
through a fibre, but the bottleneck is with the router electronics. That's
what I've gathered from this anyway:

http://www.eetimes.com/story/OEG20000926S0065.


--
Steve - www.digitalradiotech.co.uk - Digital Radio News & Info

The adoption of DAB was the most incompetent technical
decision ever made in the history of UK broadcasting:
http://www.digitalradiotech.co.uk/da...ion_of_dab.htm

DAB sounds worse than FM wrote:
Mmm. The technology used to drive it aint that good..last time I looked
it was about 8Gbps.

It probaably is something like that.

My first 64K leased line install was about 5K a year. I can get 24x7
ADSL for 30 quid a month with 10GB packet limits..

Unheard of..I think about 10K a year will by you 2Mbps leased line these
days..with internet. Used to be nearer 100K.

Bandwidth doesn't 'cost' anything really - what costs is the rails on
which the trains run as it were.

Run more trains and it gets cheaper per unit train ;-)

Adrian C wrote:
Should have used skype ;-)

DAB sounds worse than FM wrote:
In my time in that field, that was broadly true. Routers stayed the same
price, but they sort of went up by double the bandwidth every time you
blinked.




yes, and the more its used the less it costs so to speak: however the
lifetime is short, kit sort of migrates outwards as cores speeds go up,
and the old kit can't cope.

I doubt the average Cisco in an ISP's machine room stays there more than
a couple of years.

DAB sounds worse than FM wrote:


Or rather, as with PC's the price has stayed the same, but the bandwidth
has gone up and up.


Domestically, thinking back to Cliff and his 'tenner a month' service at
a rollicking 2400 baud, which with phone calls was at least 20 quid a
month if you used it at all, we are STILL at between 10 and 20 quid a
month, because that's what 'people are prepared to pay'

But we get something like 100 times more bandwidth at the absolute
minimum, and some get a lot more..

Adrian C wrote:

You forgot the cost of the billing system. This has a huge impact on the
end user costs.


If you run without a billing system, the calls can be free.

Noted.


- Nigel


--
Nigel Cliffe,
Webmaster at http://www.2mm.org.uk/

In message <Q4cL8sBzjF1HFw$H@amac.f2s.com>, Alan
<junk_reply@amac.f2s.com> writes


You are actually *able* to use 6Mbps ?

or ... if you use it ... does it become 3Kbps (like mine does?)

Mine was, in 1999, about £22 per month for 44.8Kbps

Now, the sums are more difficult due to the rate limiting
--
Rex M F Smith

Rex M F Smith wrote:

Eh ?

More like 1p a minute.


Which you use about 1% of.


I get and use it yes.


Get a better ISP then ! Mine's IDnet and damn good they are too.

Graham

The Natural Philosopher wrote:

This pdf has a figure of how router speeds have gone up over time on page
49:

http://pnrl.stanford.edu/light_flow/BCSv6.0.ppt

Pretty bloody impressive really. There's trouble ahead apparently though,
because the memory manufacturers concentrate on capacity, and memory's
bandwidth hasn't kept up.



Okay.


--
Steve - www.digitalradiotech.co.uk - Digital Radio News & Info

The adoption of DAB was the most incompetent technical
decision ever made in the history of UK broadcasting:
http://www.digitalradiotech.co.uk/da...ion_of_dab.htm