spacer 2Telebyte Gfast Testing Guide 320

gfast map nov

Dark Green: Firm commitments from incumbent: BT (10M), Belgacom, Australian NBN, Swisscom,  Austria, Bezeq Israel, Chunghwa Taiwan, Telus Canada, Telekom South Africa, SK Korea, (U.S.) AT&T, Century, Frontier, Windstream, Belgium, Omantel

Mid Green: Smaller carriers in Germany, Norway, Finland, Japan

Light Green: Incumbent likely:  France, Germany, 

Sharon-White-200?10-15% performance increase for British Telecom. Last year, I reported that G.fast could run 100 megabits faster if the frequencies below 21 MHz didn't need to be protected. British Telecom has been experimenting with reducing the carve out to increase the speed of their new offering. ASSIA now is proposing to go further, with coordinated Dynamic Spectrum Management to minimize the spectrum used by the VDSL. G.fast could then take advantage of the unused spectrum and increase the speed to the consumer. One box would coordinate all the lines, allowing faster speeds to both the VDSL & G.fast customers.
 
We don't have test data yet so the improvement can only be estimated. One estimate is 100 megabits more or 100 meters longer reach, but for now I'm being cautious and saying 50+.
The amount will vary depending on the actual lines in each binder but should be significant. ASSIA, Alcatel, and Huawei offer DSM systems that presumably could be rapidly configured for the new functionality. 
 
The kicker, as usual with vectored DSL, is that competitors must work together. Diplomacy is difficult, even though this is an obvious win-win for everyone. DSM constantly monitors the actual lines. Often, the required speeds can be achieved with lower power/different PSD mask. That in turn reduces the interference and allows other lines to run faster. 
 
DSM has been part of the standard for a decade and works well. The original DSL standard was deliberately conservative because in the 1990's, there was no reliable way to do real time measurements.  When DSM reached the field a decade later, both speed and reliability significantly improved. This was particularly important for telcos selling TV over broadband; dropouts and buffering went down considerably.
 
Current G.fast systems start at 21 MHz, to provide a margin of safety for VDSL that can run up to 17 MHz. Eliminating the VDSL - with fair concessions from the incumbent to the other operators - is the right move for better service to consumers. Once vectoring was developed, DSL became a natural monopoly. That's frightening to regulators, who prefer to rely on competition. 
 
Competition worked very well a decade ago, when 6-10 competitors drove French and British prices to half the level of the U.S. As the market eliminated all but three or four companies, the benefits of competition started to erode. In Britain the last five years, prices have gone up 25-40%. 
 
Some of the increase was due to Ed Richards at OFCOM being a nice guy but not having the courage to do what needed to be done. BT has a monopoly on lines in half the country and a friendly cable company which likes to raise prices. Richards never should have allowed BT to raise line rentals while costs went down. Sharon White looks stronger, but hasn't evolved a policy effective with weaker competition. 
 
(Structural separation of BT is not the answer. Most of the benefits will go to the other telcos, not the consumer. What Britain needs is a requirement that the carriers deliver a robust Internet at a fair price. Separation is not likely to deliver that.)
 
A thoughtful regulator would eliminate the unbundled lines as G.fast rolls out. Technology has made that obsolete. Consumers are paying the price as they are running 100 megabits slower than they should. Competition wouldn't disappear but move to a different level. ISPs could continue to compete on services, offerings, the backbone, and especially how well they treated the customer. Eliminating wire unbundling is good for consumers iff the regulator sets an appropriate price for the bitstream unbundling. Jochen Homann in Germany tried to do that but couldn't resist the political power of DT. Prices in Germany are going up by $2-$5/month.
 
Unbundling in England is going to die in a few years no matter what the regulator does. Cable will run at over 400 megabits and BT's G.fast is aiming for 300 meg. The others will be offering less than 50 meg to most. They won't be competitive. Almost certainly they will lose customers until they become uneconomic and get out of the business.
 
Holding back the sea is not easy.   
Important conflict of interest note: I'm on the Advisory Board of ASSIA and have done significant (five figure) work for them over the years.

The Site for gfast 230
 

G.fast News
A remarkable 400 people attended the very strong Broadband Forum BASE events in Berlin and Las Vegas. Trevor confirmed BT would pass the million this year. Cioffi projected “Waveguide DSL” could carry 10 gigabits a kilometer as well as a terabit 100 meters. Werner sees a 4X improvement in upstream with cDTA. Much more in next issue.

Deutsche Wants a Gigabit, Finally Realizes 50 Meg Isn't Enough http://bit.ly/2zeZ5oZ
Deutsche Telekom is finally realizing that 50 megabit DSL won't make it against gigabit cable. VP Franz Seiser is blunt. "We must change radically, become disruptive and, above all, throw away things," he proclaims at BBWF. After years of DT insisting 50 megabits is plenty, we now hear "it is about Gigabit products" from DT's Robert Soukup.  
    A lucky building in Frankfurt will receive 500+ megabit service as ultra-conservative Deutsche Telekom experiments with G.fast. Soukup told BBWF, "We're going to have a field test in Frankfurt with G.fast and Fiber To The Building (FTTB.) We will know by the end of the year if this is the right way to go." Hint to Soukup: Yes it is. G.fast is working well at a dozen telcos I;ve talked to.
     The details are surprising. DT is going for CORD, Open Source, Calix, and Radisys. http://bit.ly/2zeZ5oZ

*** The new Telebyte Guide to Testing Gfast follows the Broadband Forum IR-337 Gfast test specification, the same used by the University of New Hampshire (UNH-IOL) for Gfast certification testing. Free download http://bit.ly/telebyte (ad) It is the best technical guide to G.fast  I have seen. Grab it. Dave

1.6 Gig in Sckipio-Calix Test http://bit.ly/Calix16
A telco tells me they are getting impressive early results from the Calix 48 port DSLAM with the new Sckipio 212 MHz chips. There still is work to do but this is encouraging. 
    Carriers want DSLAMs with more than 16 ports to reduce the deployment costs from the basement or larger field cabinets. Speed matters to the marketing side of the company; AT&T's CEO believes he must offer a true gigabit to match cable. (They've been getting ~750 megabits with first generation chips. http://bit.ly/Calix16

*** Self-Healing Wi-Fi With ASSIA Real-Q 
Beyond-the-Box visibility and control extends quality-of-experience (QoE) beyond the gateway to the end-user device for every device in the home. Based on ASSIA technology, proven across 80 million subscribers http://bit.ly/2dj7FJk (ad)

Reverse Power 4 Port DSLAM for Australia http://bit.ly/NetcommRP
Australia is connecting 1M homes to G.fast, some with a Netcomm distribution point mini-DSLAM. It's a small unit designed for pole or pit mounting. It's waterproof, pressure proof, and temperature resistant. Their matching home modem is bittorrent friendly, with two USB ports for a hard drive dedicated to sharing.
     A reverse power unit at the customer, the NDD-0100-01, can save the cost of bringing power to the DSLAM. They don't expect many orders until the second half of 2018, as nbn is waiting for the second generation chips. Netcomm demonstrated RP with BT Openreach in August. http://bit.ly/NetcommRP

*** Sckipio's Three advances are taking G.fast to the next level.http://bit.ly/Sckipio (ad)

Australia Makes it Official: G.fast to Million Plus http://bit.ly/GFAussie
No news here. In September, 2015, I reported Australia's nbn Going G.fast. This June. I reported the million home fiber to the curb (kerb?) was beginning. Unfortunately, they are no closer to figuring out where to find the needed $10B to $20B to cover the cost overruns. Instead, the parties are battling in Parliament about who is to blame. http://bit.ly/GFAussie

2 Bonded 212 Lines = 3 Gigabitshttp://bit.ly/twobonded
Sckipio at BBWF is demonstrating 3 gigabits down, nearly a gigabit up, over two phone lines, bonded. Twice the bandwidth (212 MHz instead of 106 MHz) times two lines is fast. Sckipio does great demos; at CES, they showed G.fast first generation chips delivering almost 1 gig upstream.
    “Sckipio is pushing Gfast to astonishing speeds with production silicon,” CEO David Baum proclaims. Calix is using the SCK23000 chipset in their 48 port gig+ DSLAM at the show. http://bit.ly/twobonded

Read more ...