x
100G Ethernet

The Gigabit Tipping Point

We are entering the gigabit era in broadband. This jump represents an increase of 100 times over the current US average and is realistically within five years of deployment in North America.

While there is "hundredth-monkey effect" speculation about the cause, there is no doubt that Google (Nasdaq: GOOG)'s 1 Gbit/s fiber network in Kansas City and Austin tumbled the first domino. CenturyLink Inc. (NYSE: CTL) has delivered gigabit services to Omaha, and AT&T Inc. (NYSE: T) has announced it will also bring 1 Gbit/s to Austin.

Google's role in the gigabit tipping point should not be understated. Broadband rollout friction is primarily the result of higher costs in North America. The costs of network plant trend higher than in other countries because such a small percentage of the population lives in high-rise urban apartments, where tremendous efficiencies in the number of switches and repeaters to cover subscribers can be achieved.

Instead carriers must maintain a massive network of outdoor wiring that's exposed to the weather and is millions of miles long. Google limited the geographic scope of offerings in the terms and conditions, and only focused on profitable neighborhoods [so-called "fiberhoods"]. The result was a much more cost efficient rollout. Similar terms will benefit other carriers and accelerate the availability of gigabit services.

It is important to specify that we are talking about FTTH solutions, implying an overlay of the copper plant. But there is an alternative: G.fast vectoring, which can theoretically achieve 1 Gbit/s, has been proposed as a more cost-effective solution. It re-uses the final few meters of existing copper to the home. This deployment relies on a fiber to the distribution point [FTTdp] infrastructure. The cost-reduction of G.fast with FTTdp is primarily driven by not entering the subscriber home.

The second wave of gigabit access will take place with the cable providers. Already more than 80 percent of the cable installed base has DOCSIS 3 and DOCSIS 3.1, which can achieve 320 Mbit/s and theoretically up to 1 Gbit/s with channel bonding, which will be the next phase of deployment.

There has traditionally been a performance gap between wireline and wireless access. However, this may be about to change. While LTE is still fresh out of the box, several carriers have announced commercial LTE-Advanced (LTE-A) networks, such as SK Telecom (Nasdaq: SKM) in South Korea and Telstra Corp. Ltd. (ASX: TLS; NZK: TLS) in Australia. (See SKT's LTE-Advanced Subs Growing Fast and Ericsson Boasts LTE-Advanced Breakthrough.)

LTE-A can theoretically support bandwidths up to 1 Gbit/s by aggregating up to five 20-MHz component carriers. These initial deployments, however, are limited to speeds in the 150-300 Mbit/s range. It is expected that we will see LTE-A networks hit North America before the first half of 2014.

More than video and cloud
Beyond forecast models, what could you actually do with a 1 Gbit/s connection? Earlier this year, KC Digital Drive hosted a "Hacking the Gigabit City" event to explore apps that would take advantage of gigabit speeds. The dominant theme involved the delivery of cloud-based services. For instance, business software, such as CRM, Microsoft Office tools, and backup and recovery solutions could be made available through a single click.

Video was also a dominant theme, with interactive Massively Open Online Courses (MOOCs) and Tele-Health applications leading the discussion.

While video and cloud data services represent elephant-class services, however, the second opportunity for gigabit is the scale of mouse-class services. Larger pipes enable orders of magnitude: more sessions to Internet services such as newsfeeds or wellness applications, not to mention pictures of cats…

NPD Group Inc. estimates that by 2022 the average household with two teenage children will own roughly 50 Internet-connected devices. This increases the multiplier of sessions by a factor of 50 -- where each device could be running hundreds of sessions, feeding data to multiple Web-connected engines. The effect will be to make the Internet recede into the background, blurring the line between the digital and brick worlds.

It is important to clarify that a 1 Gbit/s access connection does not guarantee high-speed delivery. For instance, Google's Kansas City network delivers Netflix streaming video at an average rate of 2.55 Mbit/s, despite the gigabit access speed. The service delivery rate is governed by other Internet bottlenecks. Therefore, gigabit access will require upstream investments in the infrastructure for its full potential to be realized.

It seems that before the future even has had a chance to experiment with 100 Mbit/s, we have been thrown into a 1 Gbit/s world. The diminishing gap between wireline and wireless speeds will accelerate penetration and shape consumption of gigabit-class services faster than we may expect. Are you ready for the tipping point?

-- Vish Nandlall, Head of Strategy, Marketing, and CTO, Ericsson North America

Page 1 / 2   >   >>
derac7020 9/17/2013 | 2:20:06 PM
Re: Not yet.. Oh, I agree.   My point is that they decided to not further expand the services unlike ATT which is still pushing U-Verse.   And yes, they moved all of the non-FiOS wireline business that they could.   Interestingly, the Frontier purchase in Oregon and Wash included the FiOS deployment there.   Frontier really had no interest in it and let it die on the vine.
evisnan 9/17/2013 | 2:04:42 PM
Re: Wireless vs wired + backhaul Agree with fiber, I was using copper as a proxy for typical broadband experience. I also agree that bonding pairs will increase bandwidth, and therefore rates. Clearly there are trade offs with SNR, which limits the loop length, and therefore achievable penetration...but gigabit over copper is achievable. My point was that the increase in the spectrum for LTE-A with carrier aggregation is narrowing the gap, and that gigabit over wireless is also achievable. The question is what is practical (:
evisnan 9/17/2013 | 1:45:41 PM
Re: Not yet.. While I agree cable has an advantage, all is not lost for the humble copper. The advances is G.Fast and improved vectoring does help with the cost, in that you can run fiber from the exchange to the node, and re-use the copper from the node to the home. The end deployment requires a grab bag of approaches to resolve the best unit economics. FTTH and FTTdp, aerial versus buried, etc I think some of the techniques employed by Google can be good models for other carriers. For instance, the "fiber hood" concept forces a minimum penetration of households to sign up prior to rollout. This reduces the speculation on what the penetration rate will be, and affords the luxury of "waiting" until there is a sufficient density of subscribers prior trenching. This also reduces the number of unique visits to the neighborhood for the installs - it can be done with just a few truck rolls.
brookseven 9/17/2013 | 1:45:21 PM
Re: Wireless vs wired + backhaul evisnan,

The difference between VDSL2 and wireless in your spectrum terms is that every pair is an independent piece of 30Mhz spectrum.  Which means that as a burst to a single subscriber, they are close.  If you have lots of traffic to all subs they are not even in the same time zone.  Lets put 100 users on a cell and then realize that VDSL2 is 100x the bandwidth, if they are all using it.

Once you build out fiber, it is game over.  You have essentially unlimited spectrum to each fiber.

seven
evisnan 9/17/2013 | 1:29:30 PM
Re: Vish in Canada Back from the great outdoors, and happy for my Internet connection, thanks for moderating while I was out ! Vish
evisnan 9/17/2013 | 1:27:43 PM
Re: Wireless vs wired + backhaul I would agree it is not likely we need parity in wireline and wireless speeds, but the gap is closing. If we consider that copper speeds, by dint of penetration, define national broadband - then we are talking about a medium that is limited to < 100MHz, and in practice VDSL is 30MHz. We are approaching similar spectrum bandwidths as we aggregate LTE carriers with LTE-A. I would agree that, in practice, that per user throughput with LTE-A will not average 1Gbps. Your point about backhaul is dead on. This is where the approximate convergence of ,say, several 300 Mbps aggregated carriers and the ability to backhaul a gigabit creates a causal loop for gigabit services.
Phil_Britt 9/11/2013 | 2:56:00 PM
Re: Not yet.. I wholeheartedly agree that a few trials, no matter how successful, don't mean that gigabit ethernet will be common anytime soon.

 

Remember how many years we were at 2.5G, waiting for 3G? Granted, 3G to 4G has moved much more quickly, but it didn't require the same level of forklift infrastructure improvements as 2G to 3G.

 

Slow development of promising technology is nothing new. The coasts had cable several years before it finally became widely available here in the Midwest. Large scale broadband deployment was similarly slow in developing, though now its commonplace in most areas. I expect gigabit ethernet to be a similar long haul.
Phil_Britt 9/11/2013 | 2:55:59 PM
Re: Not yet.. I wholeheartedly agree that a few trials, no matter how successful, don't mean that gigabit ethernet will be common anytime soon.

 

Remember how many years we were at 2.5G, waiting for 3G? Granted, 3G to 4G has moved much more quickly, but it didn't require the same level of forklift infrastructure improvements as 2G to 3G.

 

Slow development of promising technology is nothing new. The coasts had cable several years before it finally became widely available here in the Midwest. Large scale broadband deployment was similarly slow in developing, though now its commonplace in most areas. I expect gigabit ethernet to be a similar long haul.
mendyk 9/11/2013 | 11:21:26 AM
Re: Vish in Canada Windowless rooms, stale sandwiches, three-hour PPT presos -- that's where the real ideas come from.
Sarah Thomas 9/11/2013 | 11:03:29 AM
Vish in Canada FYI, Vish may be a little slow in responding as he is in an off-site management in Canada, riding horses and camping out in tents. Now you know where all of Ericsson's big ideas come from...
Page 1 / 2   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE