[LINK] NBN to be 1Gbps

Craig Sanders cas at taz.net.au
Thu Aug 12 16:46:34 AEST 2010


On Thu, Aug 12, 2010 at 02:10:54PM +1000, Philip Argy wrote:
> What practical use is 1Gb/s for the majority of people right now?

depends on whether you think enabling and encouraging LOCAL content
creation - both professional and amateur, commercial and personal - is a
good thing or not.

the internet is, remember, a peer-to-peer network, not a producer-
to-passive-consumer transmission.

there are many applications that would become possible with faster
networking.  many that have already been thought of but are unfeasible
now, and many more that haven't even been thought of yet.


<SPECULATION TYPE="VAGUE LOW-DETAIL HANDWAVING EXAMPLES">

with that sort of bandwidth, even with 100Mbps, it is feasible,
for example, for individuals to put up family photo albums or do
live-streaming of important family events like birthdays, weddings or
whatever direct from a digital video camera to those who are unable to
physically attend(*).


another use is for on-line access to off-site backup facilities capable
of storing more than spreadsheet and text files. current bandwidth
speedds and especially the asymmetric nature of adsl does not allow
backup with this kind of volume. the backup facilities could be provided
commercially, or informally by family and friends doing favours for each
other ("sure, you can store a backup of your home movies and stuff on my
little home file-server. is 100TB enough to start with? bring another
disk and some beers next time you come over.")

ditto for off-site storage of security camera feeds. burglars can't
smash the hard drives holding a recording if it's not on the premises.

btw, if my Q&D calculations are correct, 100TB would take approx 10 days
flat out at 1Gbps (ignoring protocol overhead etc). fortunately by the
time we are able to talk so blithely of 100TB of storage, we'll be using
10 or 100Gbps fibre.  fibre has an upgrade path.  copper doesn't.


yet another is a distributed proxy/cache network using a
bit-torrent-like protocol. it would really only be useful for static
data, but could effectively be a huge akamai-style CDN for every site
accessed via the distributed proxy. the more downloaders there are, the
more uploaders to support them.  AFAIK, nothing like this exists yet -
partly because, with ADSL, upload speeds are far smaller than download
speeds.

it's something that could be implemented just by integrating bit torrent
into the browser, and a bit of smarts on the web server to check
something like the current HTTP Accept-Encoding: header and serve a
torrent file rather than the actual file if the client supports it.

(and there are probably numerous other things that could benefit from
this kind of "self-organising" automatically adaptive network protocol)


and one other really obvious use is for hospitals and universities and
research institutes to have direct high-speed links to each other (such
links could be used by aarnet, or used as a supplement to aarnet) -
e.g. every australian university could have a direct 1Gbps link to the
synchrotron and other facilities that generate VAST quantities of data.
medical students could have access to live (and recorded) operations and
other procedures performed by the top specialists in the country, rather
than just that are close enough to physically attend.

</SPECULATION>



(*) this would obviously require a revamping of current ISP BW quotas.
at minimum, traffic within australia (i.e. either within an ISP's
network OR carried over the NBN from one NBN-connected node to another)
should not be counted as part of a user's quota.


> Whilst I personally think we should be aiming for 100 Gb/s in the next
> 5 to 10 years to cultivate innovation in an ultra high speed bandwidth
> environment, it's a bit of a con job to entice people with 1 Gb/s to
> the home when most households, if they have anything, will have Cat 5
> cabling and components which won't support gigabit Ethernet no matter
> what they do.

networking hardware upgrades will be needed, obviously. do you really
think that a few dollars for a cat-6 or better cable is going to matter
much compared to a few hundred for a new router / "modem"?


as for existing in-home cabling: given that the difference in price
between cat-6 and cat-5 cabling has been negligible for YEARS, and the
fact that most of the expense is paying for a licensed tradesperson
to actually install the cable, most of us who did so(**) paid the
few extra dollars to get cat-6 cabling(***).

in any case, as with current telephone and broadband connections,
everything past the network boundary is the responsibility of the
customer, not the service provider.


(**) almost all of whom are geeks, because nobody else cared about or
wanted network cabling in their homes.

(***) and many of us will be looking to upgrade our home networks to use
fibre rather than copper as soon as it's economically feasible. at the
moment, fibre NICs and switches are too expensive for home use.


> Surely it's better to get wider coverage at a reliable base speed
> of 12Mb/s in the next couple of years than to give a few Tasmanian
> households a 1Gb/s fibre link to their front door?  

1. NBN stands for National Broadband Network. last i checked, the nation
consists of more than just Tasmania. certainly more than just a few
households there.

2. No. spending 6 billion dollars to patch up some of the gaps in
the current network (and to maintain the current near-monopoly) is a
short-sighted waste of money. even if it's 100% successful, it'll still
only improve things for a small number of people (and even they'll just
be a small subset of people who have inadequate service now).


FTTH on the other hand will more than cope with any reasonable (and many
unreasonable) needs for the forseeable future, AND offer an upgrade path
as signalling lasers and fibre switches get better and faster.

and, presumably (i'm no expert on the topic - but i do know enough to
understand the qualitative rather than merely quantitative difference
between eletrical and optical comms), there's also significant room for
improvement in manufacturing processes to produce higher-quality optic
fibre cables than we're currently capable of.

once the basic infrastructure is in place, there are many possibilities
for incremental improvements.


> And where's the commitment to an accelerated IPv6 implementation that
> is more critical than bandwidth.  Not much good having the fibre at
> the door but no spare IP addresses for what's in the building to
> connect!

IPv6 is definitely something that australian ISPs should be
implementing. most aren't bothering, but some already are.

but that's really got nothing to do with whether the "local loop" is
FTTH or xDSL.  

If anything, moving to IPv6 will be easier with the NBN as one single
policy change can start the entire country switching over to IPv6. and
won't get bogged down in disputes and strategic delays by ISPs.


> Very shallow debate I'm afraid.  Let's propose to give every man,
> woman, child and newborn baby a free top-of-the-range iPhone 4 @ $800
> - 22 million of them would cost less than $18 billion so it puts into
> perspective the size of the $43 billion or possibly more for the NBN
> Co's fibre.

yeah, very shallow. it makes perfect sense to compare disposable
consumer gadgets to national infrastructure. the iphone4 will be
obsolete within 6 months, and most of them will be disposed of within
2-3 years (a small percentage will be handed-down, but almost nobody
wants 3 year-old IT stuff now and there's no reason to believe that
people will be any different in a few years)

craig


ps: minor technical quibbles aside, the ONLY real problem i have with
the current NBN plan is with the idea that it will, or even should,
be privatised 5 years after it's fully operational.  FFS! part of the
reason for the NBN is to escape the telecommunications disaster that
resulted from privatising telstra!

-- 
craig sanders <cas at taz.net.au>



More information about the Link mailing list