Stephen,

Where did I ever say it was impossible...useless on the other hand is my commentary.

seven

Wow, what if DSL Rings works fine with the bandwidth that we say, & major European telcos have verified, Nyquist & Shannon's principles are not shattered, and Seven is WRONG...?

I think we should be worried about Charlatans that masquerade as know-it-alls.  I was going to compliment LightReading for their amazing new spell-checker that made BobbyMax appear as brookseven and make his drivel more readable and less grammatically repulsive, but instead I'll just bookmark this post for its volume of petard-hoisting content.  I especially like the total inconsistency in the early comment about "the cost to lay down fiber is high" and then "you can not assume much more than a 25% take rate to ammortize..."  You do realize that this blows the whole business case for FTTx out of the water...?

Perhaps Tellabs regularly calls major European telcos "liars & cheats"?  Do you find this works for you Seven?

Steve.

Seven, I can't disagree with the bulk of your disertation, and enjoy reading the back and forth. 

I especially agree with this qualifier: "if you have to lay copper to use it - it is better to lay fiber.  The cost of laying is the same and since the electronics on the fiber side is cheaper then bonded DSL is the wrong choice."

We are faced with the decision most of the time, as enablers and developers, to choose between making the fast quick buck using what is onsite or deploying fiber infrastructure that will be future proof for the next 50 or so years.  My hope is that we defer to the latter wherever possible, as I think it strategic to our nation and economy's future..

And, until we get to deploy something like quantum entanglement based broadband connections to surpass fiber someday (no wireless spectrum!), we'll have to limp along in some areas with copper and wireless, and the consequent bandwidth and congestion limitations unless we opt to proactively deal with them ahead of time.

.

So, no actually I am greatly concerned about the Charlatans that hang around the DSL industry.  They cause people and companies to invest good money on dead end paths.  My best example of this was an early Ikanos slide that had two facts on it - VDSL can do 100 Mb/s and it can do 9 Kft.  People (meaning Telus' CTO office) took this to mean "at the same time".  Of course, it didn't and people wasted a whole bunch of time figuring that out.  Our friends in the carriers had to force the DSL forum to make rate versus reach objectives (thanks to what is now AT&T).  Equipment makers and chipset folks whined but it got passed and things greatly improved.  I plan to put up a complete post here to illustrate the problems.

Fiber, DSL, Wireless and Cable are all propogations of electromagnetic energy down waveguides.  They all follow the same rules.  These rules as it comes to communications are governed in terms of bandwidth by information theory.  Shannon and Nyquist a long time ago told us that this in the end is governed by Signal to Noise Ratio.  The US Military invested heavily in modulation theory and created most of the schemes that we use today.  For example, I worked on leased line modems that used DQPSK in the 1970s.  This is the exactly the same technology being discussed as new in the fiber world.  It is not new.  What is new the technology which allows us to create modulators and demodulators at much lower cost and much higher performance than we used to.  There is still work going on in this stuff.  You talked about DSM L3 and then there is the work on MIMO.  All of these things help with noise reduction to improve SNR and thus have a bigger communications channel.  The thing is that the big gains technologically were made a long time ago.  Now if we get 10% improvement, that is a good thing.  

What about this DSL bonding stuff and why is it different than say running ML-PPP across all the pairs?  Since most of the time two or more pairs running to the same endpoint run in the same copper bundles you can reduce the noise coming by training the system.  When the system trains the pairs the receivers can estimate the noise from the other transmitters and attempt to remove it from the transmissions when the training period is over.  There is some gain in SNR from this effect.  Of course, none of the inherent noise in the system is removed and transitory effects (like rainwater in the cable) sometimes make things worse.  But be that as it may.  The technology basically works and is DSL is capable of delivering moderate speeds.  Of course, good clean copper in abundance is required.  Of course IMA, ML-PPP, or bonded T-1 (like we used to do in the old days for video conferencing) are all just slightly worse and require no new transmission infrastructure.

Now, higher rates over copper is attractive to carriers.  The cost to lay down fiber is high.  The problem is that copper is basically limited by its distance.  The wire itself asks as a low pass filter that cuts down the frequencies that will be supported over the wire.  So, this 30A profile is limited to about 100M.  Not bad really, but that means that if you are in rural America...the rings don't work.  There are not multiple houses that close together.  In fact, the 30A profile is designed for in-building for apartment buildings.  Lower speed profiles can reach longer of course, but then the bandwidth is greatly reduced.

So, what we have been doing in the DSL world is moving the DSLAM port closer to the home.  AFCs first paid GA DSL port served a customer 22 miles from the CO.  The DSLAM port was essentially remoted over fiber and so the DSL link itself was plug standard. So, when we talk about this remember the distances involved and remember that there are multiple DSL modems on each end of these circuits.  These are not free and also recall the price of an indoor ONT (like the ones used in NTT's deployment) are close to the cost of a DSL modem.  So, the electronics cost of bonded DSL is higher than that of fiber.  The big cost (and theoretically time) advantage is around the cost of deploying the actual fiber.  Of course, few have made FTTC work economically.  Those that have did so in the same areas that FTTH has worked economically.  

Also recall, that these very high speed services are not for everyone.  Take rates are not 100% and so when you deploy these services you have to account for the fact that many folks will not buy them. So you run one of these nodes out at a distribution box (which you then have to power which is a cost), and you can not assume much more than a 25% take rate to ammortize the hub of the DSL ring system.  Which pretty much makes the costing problematic.  But that is not the biggest issue.

To introduce a new technology costs a carrier $20M+ or so they have told me.  So, it is NOT a site by site choice on the technology as a Tier 1 carrier will need a whole lot of sites to make a technology worth approving.  That is the issue for these technologies.  They simply do not have the number of sites that they can meet to cross that threshold.  A Tier 1 carrier needs a technology that can use in all cases where it is going to do something.  Fiber meets that as do some single pair DSL choices.  Bonded DSL does not, because if you have to lay copper to use it - it is better to lay fiber.  The cost of laying is the same and since the electronics on the fiber side is cheaper then bonded DSL is the wrong choice.

Does bonded DSL have some applications?  Yes, it can in 10 Mb/s Ethernet circuits to small business customers.  There are some smaller carriers looking at bonded DSL for IPTV deployment over longer reach.  As a competitor of FTTH or DOCSIS 3.0?  Nope, DSL is done on that front.

And Telsa - no meatheads - just liars and cheats.

seven

Sounds like you meatheads just like to argue and torque each other...

Kinda feels like we have a sales guy vs. engineer thing going too...

...which is fine because not only is it educational but entertaining as well.

:)

You are each right....from a certain point of view.

Seven has the need for speed and future proof bandwidth...so fiber is the obvious choice, especially since he's willing to pay for it. 

Steve speaks to applications where when you have adequate copper available, fiber like performance can be had for a very reasonable price and for site specific reasons...such as what was done here:

http://www.indianikah.com/a746064-sacred-wind-positron-occam-expand.cfm

The choice to go with fiber or copper has just as much to do with cost as with site specific conditions, and the distances, number of pairs available or not, etc.

One would choose fiber right off if it was obvious there was just not enough enough copper available or you had a greenfield situation today (and stim $$)

And if you had ample copper available and moderate to close proximity to your heaviest users or towers?  You'd have to be a drunk and rich to want to lay new fiber...(or have stim $$ and a very strategic plan).

There are situations where a complete fiber buildout is appropriate though:

http://www.lightreading.com/document.asp?doc_id=194295

And there are also applications where you really don't want to hamstring your future bandwidth, and choose between even 2 configurations of fiber (GPON vs. FTTH), like they did here:

http://4g-wirelessevolution.tmcnet.com/broadband-stimulus/topics/broadband-stimulus/articles/88647-occam-rapid-systems-deliver-broadband-central-florida.htm

Bottom line is you can both be right...and it is Friday.

"Your DSL rings did not make any sense when you pitched them to me at AFC..." = whole new ball game.

Note - we are talking about CELL SITES!

Seven,

You like to think you remember how a technology works when you slander it but the truth is that we use bonded copper (typically 24-pairs of ADSL2+) to the pedestal (or Distribution Point) and then we use VDSL2 over the 500ft or less to link the homes together via home gateway nodes in each house.  The bonded link can also use vectoring if the DSLAM supports it across that many pairs.  On the VDSL2 side the transmission curve re-starts at each house so we can achieve what the European telcos have witnessed.

Spare copper is not required as the houses on the ring donate their backhaul (ped to CO) pairs to the bonded link.  Any spare copper would be included in the backhaul bond to increase available bandwidth at the ped - so it would help, but is not required.  We add QoS (traffic prioritization) so cell traffic can be prioritized above your WoW game or the telco can sell you an SLA that gives you better performance on your games as well.

Yup, you're still wrong.

Steve.

No you are playing the normal bait and switch game.

Your DSL ring requiring a 30A profile does NOT do 6Kft in a single hop.  That is the old DSL bait and switch.  My favorite was the first Ikanos pitch was that it was a 9Kft and 100Mb/s technology.  Not at the same time of course.

Which is my point here....You can run all kinds of bandwidth at very short loop lengths over lots and lots of copper.  None of which has anything to do with the actual state of the network. 

Copper is good for what it is good at.  Heck there are Ethernet Phys that run it to Gigs and backplane drivers doing 10s of Gigs.

But let me remind you of your assumption.  You have lots of free and clean copper running to a cell site.  What the operators tell me is that this is an occasional thing and they might consider bonded copper in those specific cases.  In the vast majority of cases, they end up having to do construction.  Once you are there, they are not going to plow in brand new copper.  That is the basic issue. 

Your assumption - EVERYWHERE has LOTS of free copper.  Reality - FEW places have LOTS of free copper.

Remember, you business case only works if you can avoid laying cable.  The cost to lay fiber is about the same as it is to lay copper.  So, if you have to lay anything - you are done.

seven

Seven,

You mentioned that our ring didn't make sense.  I provided data that showed that it provides significant bandwidth in telco testing that means DSL Rings absolutely makes sense.  Hurts to be wrong doesn't it?

Steve.