On funsec we have had a discussion on broadband providers and bandwidth
limitations, pretty much what we rehearsed here.
Michael brought up an interesting case from a decade ago, which speaks of
some litigation issues we did not discuss. It is also
interesting to hear his view as a client on "been there done
that". Interesting reading.
Gadi.
Critical mass is approaching. There's only so long that North
American consumers can be held back from bandwidth-hogging
applications and downloads while parts of the world have long since
upgraded to 10Mbit/s bidirectional (and beyond) consumer-grade access
speeds.
Both cable and DSL providers are about to have a very loud wake-up
call, and from here, I see absolutely zero uptake of newer technology
and infrastructure to offset the inevitable.
not that I'm arguing (really) but what wakeup call? where is the
competition pushing to provide 'better' than the local telco/cable-op is
providing? what business drivers are there to put more bits on the wire to
the end user?
-Chris
768 ain't broadband. Buy Cisco, Alcatel, and Akamai stock!
BitTorrent.
;>
And on-demand DVR-type things which I believe will grow in popularity. Of course, most of those are overlays which the SPs themselves don't offer; when they wish to do so, it'll become an issue, IMHO.
Alexander Harrowell wrote:
Critical mass is approaching. There's only so long that North
American consumers can be held back from bandwidth-hogging
applications and downloads while parts of the world have long since
upgraded to 10Mbit/s bidirectional (and beyond) consumer-grade access
speeds.Both cable and DSL providers are about to have a very loud wake-up
call, and from here, I see absolutely zero uptake of newer technology
and infrastructure to offset the inevitable.768 ain't broadband. Buy Cisco, Alcatel, and Akamai stock!
It certainly is - just ask the CALEA folks.... and as for who is pushing the bandwidth curve, for the most part it seems to be gamers in search of the ever shrinking ping time. I suspect they make up most of our >1536kb/sec download customers.
What "parts of the world" have long since upgraded to those speeds - and how do they compare size-wise to the USA? We've got an awful lot of legacy infrastructure that would need to be overcome.
I will happily agree that it would be nice to have higher upload speeds than DSL generally provides nowadays. What are cable upload speeds like?
> what business drivers are there to put more bits on the wire to
> the end user?BitTorrent.
which uses all available bandwidth on the user link, and can/does play
nicely with other user apps... It's not a reason for $TELCO to want to add
more BW to your link though.
I suppose what I was asking is: Is there a better/faster/cheaper
alternative to your 2 incumbant solutions $TELCO || $CABLECO ?
If there were then I bet $TELCO || $CABLECO would drop prices and speed up
links... since there isn't I think we're all lucky we're not still using a
110baud coupler modem 
;>
And on-demand DVR-type things which I believe will grow in
popularity. Of course, most of those are overlays which the SPs
themselves don't offer; when they wish to do so, it'll become an
issue, IMHO.
again, these are user apps that depend on the higher BW available, they
don't drive the business to change, really. It seems to me that currently
the DVR/on-demand folks are basically walking the ledge hoping that as
they bring new features the telco's/cableco's will play nice and add
bandwidth to make these services 'work'... That might not last, there
certainly is no real reason that $TELCO || $CABLECO would be driven to
change, aside from 'goodness of their hearts' or 'hey maybe we want to
increase BW so we can offer a spiffy DVR-ish thing to our customers and
get more revenue on our flagging last-mile circuits?'
-Chris
[...]
what business drivers are there to put more bits on the wire to the end
user?BitTorrent.
The download speed is however limited by the upload speed of the peers,
which acts as its own rate-limit given that the bandwidth on broadband
connections is somewhat asymmetric.
Alexander Harrowell wrote:
>>Both cable and DSL providers are about to have a very loud wake-up
>>call, and from here, I see absolutely zero uptake of newer technology
>>and infrastructure to offset the inevitable.
>
>768 ain't broadband. Buy Cisco, Alcatel, and Akamai stock!
I'd agree wit this.
It certainly is - just ask the CALEA folks.... and as for who is pushing
And this. (sigh).
the bandwidth curve, for the most part it seems to be gamers in search
of the ever shrinking ping time. I suspect they make up most of our
>1536kb/sec download customers.What "parts of the world" have long since upgraded to those speeds - and
how do they compare size-wise to the USA? We've got an awful lot of
legacy infrastructure that would need to be overcome.
We have a lot more physical distance to cover which basically
requires fiber to get a reasonable distance from it. Even with fancy
($300-1k) LRE/dsl extenders, your limits are somewhere around 7km. Not
exactly something you can expect some consumer to jump on the costs of.
They expect the service to cost less than the computer they are going
to attach.
I will happily agree that it would be nice to have higher upload speeds
than DSL generally provides nowadays. What are cable upload speeds like?
I think that with the current market environment the only
choices will become some sort of municipal fiber builds (most people can
accept the cost via their property taxes or other means that may even be
tax deductable for them personally) or possible regulation of delivery of
"internet" services in the same way that delivery of POTS services are
necessary. Last time I talked to someone at my PUC, he was griping
about the lack of POTS services in the state. Even with all the USF and
other monies, tarrifed POTS services are not available. I think this
says something.
Me? I see a resurgance (as long as regulatory - CALEA & other costs)
in the local SPs coming as opposed to the cable/dsl cartels. Most folks
are thinking wireless these days, but I suspect that once they realize that
the cost of putting fiber across their property is actually low enough, a
number of these local isps will negotiate their own cabling paths. This
may also have some problems as if they're "good enough" we may see the telcos
and cable co's vying for access to their facilities to deliver tv/voice/data
as well over it.
Either way, the challenges in this space in the coming years at the
high end (100G and faster) as well as how to deliver content at a
reasonable speed to the end-user networks will make for a fun time
in networking.
- Jared
Jeff Shultz wrote:
Alexander Harrowell wrote:
768 ain't broadband. Buy Cisco, Alcatel, and Akamai stock!
If you don't like it, you can always return to dialup.
It certainly is - just ask the CALEA folks.... and as for who is pushing the bandwidth curve, for the most part it seems to be gamers in search of the ever shrinking ping time. I suspect they make up most of our >1536kb/sec download customers.
Gamers don't really need much in bandwidth. They need the low ping times, so they *must* ensure that there is no saturation or routing overhead. Granted, there are some games that are bandwidth intensive, but everyone's busy playing WoW. Gamers are great for detecting those really hard to spot problems that only effect gaming and voip.
What "parts of the world" have long since upgraded to those speeds - and how do they compare size-wise to the USA? We've got an awful lot of legacy infrastructure that would need to be overcome.
Japan has, for one. Definitely a size difference. In US metropolitan areas we are seeing a lot more fiber to the home. The cost will never be justified in US rural areas. Just look at Oklahoma. Most connectivity in Oklahoma will actually be from Dallas or Kansas City.
I will happily agree that it would be nice to have higher upload speeds than DSL generally provides nowadays. What are cable upload speeds like?
I would like to blame the idiots that decided that of the signal range to be used on copper for dsl, only a certain amount would be dedicated to upload instead of negotiating. What on earth do I want to do with 24Mb down and 1Mb up? Can't I have 12 and 12? Someone please tell me there's a valid reason why the download range couldn't be variable and negotiated and that's it's completely impossible for one to have 20Mb up and 1.5 Mb down.
Jack Bates
Data point: a "considerable" number of mobile ops worldwide are
pulling fibre to their Node-Bs or at least their RNCs. (No, wireline
types - not Republican National Committees, Radio Network Controllers
- you have one for every 10-15 Node-Bs, for a very rough idea)
Sources say the triggering event is the enablement of HSDPA (and
presumably Revision A for the CDMA world, although I haven't heard of
a CDMA carrier fibreing up yet). Some deployments so far have been up
to 2,000 cell sites with fibre backhaul.
So long as most torrent clients are used to share content illicitly, that doesn't sound like much of a business driver for the DSL/CATV ISP. And so long as the average user doesn't have an alternative provider which gives better torrent sharing capabilities, there doesn't seem to be much of a risk of churn because of being torrent-unfriendly.
Building high-capacity access to the home is sooner or later going to involve fibre, which is going to necessitate truck roll and digging. There's a high cost associated with that, which means there's a significant competitive disadvantage to anybody doing it in order to compete with DSL/CATV folks whose last mile costs are sunk and were paid for long ago. Residential customers are notoriously price-sensitive and low-yield.
Pressure seems like it could come from either or both of two directions: there could be some new market shift which entices customers to pay substantially more for increased performance, and to do so in great numbers, to make it cost-effective for a green-fields entrant to deploy a new network, or the cost of digging up the streets could become much lower.
Given that there's only so much TV one household can realistically download and watch per day, and since that amount of TV demonstrably fits within DSL- and cable-sized pipes already, I don't see the average neighbourhood throwing money around in order to get fibre to the home. On the contrary, here at least I see people switching providers in order to take advantage of bundles of phone/TV/cell which will save them $10 per month.
Perhaps city planners have a role to play here. In cities where the streets are routinely dug up every spring as soon as the last snow disappears, for example, municipalities could choose to invest in equal-access conduit to reduce the cost for anybody who wants to blow fibre down them in the future. Such approaches are somewhat common in the business core, but perhaps not so much in residential areas.
Joe
"Ideally" that's how it's supposed to work, but isn't how it works as of
present-day. Speaking solely about the BitTorrent protocol, upstream
does not affect downstream speed. In fact, there's a BitTorrent client
out there which specifically *does not* share any of the data being
downloaded (thus acting as a pure leeching client):
Someone please tell me there's a valid reason
why the
download range couldn't be variable and negotiated
There are several valid reasons, but with newer modulations more
bandwidth upstream is more and more of a reality. Now if we could
just turn off ISDN and POTS (and other random crazy PTT legacy)
we'd have tons more! Copper has a long way to go bandwidth wise.
Regards,
Neil.
er, that's why I put a smiley below it. Like this:
;>
In all seriousness, DVR-on-demand type services offered by the SPs themselves would be one driver. Right now, they're all overlay networks which the SPs don't view as being directly monetizable. If/when they offer such services themselves, however, I, predict this will change.
>
>
>
> > what business drivers are there to put more bits on the wire to
> > the end user?
>
> BitTorrent.which uses all available bandwidth on the user link, and can/does play
nicely with other user apps... It's not a reason for $TELCO to want to add
more BW to your link though.I suppose what I was asking is: Is there a better/faster/cheaper
alternative to your 2 incumbant solutions $TELCO || $CABLECO ?If there were then I bet $TELCO || $CABLECO would drop prices and speed up
links... since there isn't I think we're all lucky we're not still using a
110baud coupler modem
This is part of the "perfect storm" puzzle (basically,
"access monopolies are weakened or cease to exist"). See
http://www.1-4-5.net/~dmm/talks/apricot2007/perfect_storm
for the most recent incarnation of this stuff. Long story
short is that this (the whole situation with access
networks) is perhaps the most controversial/weakest part
of the story.
> And on-demand DVR-type things which I believe will grow in
> popularity. Of course, most of those are overlays which the SPs
> themselves don't offer; when they wish to do so, it'll become an
> issue, IMHO.again, these are user apps that depend on the higher BW available, they
don't drive the business to change, really. It seems to me that currently
the DVR/on-demand folks are basically walking the ledge hoping that as
they bring new features the telco's/cableco's will play nice and add
bandwidth to make these services 'work'... That might not last, there
certainly is no real reason that $TELCO || $CABLECO would be driven to
change, aside from 'goodness of their hearts' or 'hey maybe we want to
increase BW so we can offer a spiffy DVR-ish thing to our customers and
get more revenue on our flagging last-mile circuits?'
Its hard to say. There's a relatively new (well, last
Feb) paper by David Levinson and Andrew Odlyzko entitled
"Too expensive to meter: The influence of transaction
costs in transportation and communication" [0] that tries
to use economic theory and some historical perspective
(in particular, on the funding and congestion models for
roads) shed some light on this. Its worth reading as it
gives some insight as to where all of this may be going,
but as usual, its a cloudy crystal ball.
--dmm
[0] http://www.dtc.umn.edu/~odlyzko/doc/metering-expensive.pdf
OK, what drove the improvement from the 110 baud backwater to today's US
backwater? And what evidence is there that the same driver won't continue
to push?
Joe Abley
[...]
"Ideally" that's how it's supposed to work, but isn't how it works as of
present-day. Speaking solely about the BitTorrent protocol, upstream does
not affect downstream speed. In fact, there's a BitTorrent client out
there which specifically *does not* share any of the data being downloaded
(thus acting as a pure leeching client):
Yes, but if *everybody* did that, nobody would be uploading and thus there
would be nothing being downloaded.
Jack Bates wrote:
Jeff Shultz wrote:
Alexander Harrowell wrote:
768 ain't broadband. Buy Cisco, Alcatel, and Akamai stock!
If you don't like it, you can always return to dialup.
It certainly is - just ask the CALEA folks.... and as for who is
pushing the bandwidth curve, for the most part it seems to be gamers
in search of the ever shrinking ping time. I suspect they make up most
of our >1536kb/sec download customers.Gamers don't really need much in bandwidth. They need the low ping
times, so they *must* ensure that there is no saturation or routing
overhead. Granted, there are some games that are bandwidth intensive,
but everyone's busy playing WoW. Gamers are great for detecting those
really hard to spot problems that only effect gaming and voip.
You do need a high symbol rate because otherwise the cost of putting the
next packet on the wire is itself an intolerable delay. you can only put
a 1500 byte packet on 256Kb/s dsl every 47ms or so. at 1.5Mb/s it's
every 8ms at 22Mb/s it's one every .5ms...
People pay proportionality more to get semi-deterministic low-latency.
unfortunately there aren't a low of products offered specifically cater
to that market. You get your choice of 8/768 cable 6/768 dsl or maybe
fios if you happen to be in the right market.
What "parts of the world" have long since upgraded to those speeds -
and how do they compare size-wise to the USA? We've got an awful lot
of legacy infrastructure that would need to be overcome.Japan has, for one. Definitely a size difference. In US metropolitan
areas we are seeing a lot more fiber to the home. The cost will never be
justified in US rural areas. Just look at Oklahoma. Most connectivity in
Oklahoma will actually be from Dallas or Kansas City.I will happily agree that it would be nice to have higher upload
speeds than DSL generally provides nowadays. What are cable upload
speeds like?I would like to blame the idiots that decided that of the signal range
to be used on copper for dsl, only a certain amount would be dedicated
to upload instead of negotiating. What on earth do I want to do with
24Mb down and 1Mb up? Can't I have 12 and 12? Someone please tell me
there's a valid reason why the download range couldn't be variable and
negotiated and that's it's completely impossible for one to have 20Mb up
and 1.5 Mb down.
VDSL2 ITU G.993.2 supports variable and symmetric negotiation of rates.
obviously distance is a factor, cause you're down to ~50Mb/s at 1000
meters.
at&t and bell south, now at&t and at&t had vdsl rollouts that could in
theory be upgraded to vdsl2.
If you were in helsinki, I know P�ij�t-H�meen Puhelin (php.fi) would
sell you 100/24 vdsl2 for around 80euro a month.