In the quest to prepare for DOCSIS 3.0 without undertaking the necessary step of replacing aging coax with fiber, Comcast has been playing around with several solutions designed to postpone the inevitable and squeeze more bandwidth from their copper turnip. The end result? Freeing up anywhere from 25% to 50% of their available bandwidth on the coax last mile.
One of the peskier problems with their bandwidth crunch is the FCC mandate to carry analog signals until 2012, well past the cutoff date for over-the-air analog signals. Analog signals, despite being lower quality, take up significantly more space than their digital counterparts and require transmitting a channel at least twice. RCN has already started a move towards 100% digital transmission by handing out digital-to-analog converter boxes to the hold-outs. Now Comcast is also getting in on it in a move to free up at at least 250MHz of bandwidth or about a quarter of what's available. It's a much better strategy than moving channels to digital tiers to try and force customers to upgrade.
They're also looking at expanding trials of switched digital video or SDV. The current ages-old method is to broadcast all available channels at the same time, an inefficient way to do things when only a handful of cutomers (if any) are watching ESPN 473. SDV helps solve some bandwidth crunch by only transmitting the channels being watched. Unfortunately, this leaves a lot of CableCARD users out in the cold as promised upgrades to support SDV (particularly on TiVos) have failed to materialize. Don't go claiming any conspiracy theories just yet; it's probably just a sign of CableCARD's poor acceptance in the marketplace. Sony is trying to rectify that by working on a new two-way standard from CableLabs.
Comcast is also working hard to transition the back end from MPEG-2 to the significantly more efficient MPEG-4. This also benefits customers of Comcast Media Center which reportedly includes Mstar. The biggest roadblock will be upgrading legions of older MPEG-2 capable STBs to the newer CODEC, many of which aren't flash-upgradable. While the focus is going to be on VOD and premium channels, more widespread adoption of MPEG-4 could cut the bandwidth footprint of video in half.
Comcast can't get these bandwidth savers in place fast enough. The competition for HD channels is fierce with Dish planning on hitting 100 HD channels by the end of next month. Both CableVision and Verizon are aggressively pursuing new HD channels to stay competitive with satellite providers and Comcast won't be sitting still either. With so many HD sets now in place and the digital cutoff rapidly approaching, the demand for HD content is exploding.
It's also worth noting that Comcast needs to clear the way for DOCSIS 3.0 deployments. SDV is being rolled out first in markets most likely to see DOCSIS 3.0 first such as the Minneapolis/St. Paul area. DSL Reports notes that as storage gets cheaper, demand for bandwidth to access that storage anywhere will only grow. Now is the time to get ahead of the curve.
It’s hard to read through this post, because it doesn’t seem like you have a strong sense of how a cable network operates. In addition, there’s an assumption that 100% digital fiber-to-the-home is vastly superior.
With “Aging Copper,” are you talking about the line from the curb to the home? When people speak of “copper wires,” they usually mean the phone lines. Are you talking about coax?
You write about freeing up bandwidth “on the coax last mile,” as if the fiber trunkline has more bandwidth than the coax to the home. This is an inaccurate reading of cable’s bandwidth constraints.
You mention “upgrades to support SDV.” People with one-way CableCARD-supported devices are cut off from SDV, by the solution is the tuning adapters that are being developed by the set-top box manufacturers.
You mention Sony, but a number of other manufacturers has also signed on to develop tru2way TV sets.
Finally, you suggest that SDV is being deployed in advance of DOCSIS 3.0. I’ve been following both of these stories and I don’t have any sense these two things are lining up like that. Did Comcast deploy SDV in the Twin Cities? I don’t remember reading that.
In my mind, copper is copper be it twisted pair or coax. The flavor of copper being used doesn’t matter all that much since you’ll still run into the electrical limitations of the medium.
HFC is just another FTTN solution but with coax instead of twisted pair. We’ve already seen how FTTN solutions from Qwest and AT&T fail to deliver next-generation speeds. I think it’s perfectly valid to criticize this solution as entirely inadequate based on how other FTTN projects are performing and the large number of subscribers that get put onto a single node.
One of the articles I linked showed that SDV is being expanded in the Twin Cities. Comcast has also separately mentioned that they’re testing DOCSIS 3.0 in that area. Given the bandwidth constraints cable operators are already operating on with DOCSIS 1.1, I just put two and two together.
“In addition, there’s an assumption that 100% digital fiber-to-the-home is vastly superior.”
It is not just the fiber that is better is it the network topology.
In a cable system the coax does generally have alot of bandwidth compared to the twisted pair phone network, but that bandwidth is shared. When the cable networks where first installed getting as many homes as possible into the same coax ring was a good way to maximize profit. Originally the network was 1way it only sent tv channels no data was sent back, And all the connected points consumed the same data(tv channels). docsis 3.0 will have 150meg to 600meg of total bandwidth depending on the quality of the local cable network in utah it will be 150meg almost everywhere(550mhz cable systems here) in bigger places where they have newer systems or its a more profitable market they may have 800mhz-1ghz cable rings which can provide the faster speed. eather way that bandwidth is shared among every connected point on the network in many cable rings over 1500 homes.
In a phone network the twisted pair generally does not have as much bandwidth but their is a dedicated wire to each home because the telephone system is used for 2way communications. DSL relies on this setup. The phone system never needed very much bandwidth to each connected point for voice service. The fact that their squeezing multi meg connections out of old copper lines on the telephone system is actually pretty amazing.
The copper needs to be pulled out of the ground and sold for scrap. It will never handle the future bandwidth needs of American business and consumer uses.
Aren’t the fiber systems also using shared bandwidth? On iProvo or Utopia each home has a 100mb/sec. connection from the node to the home, but for every 10 homes (1000mb or 1GB) the networks don’t have a GB in the backbone do they? That would require 10Gb for 100 homes and 7TB for Utopias 7000 customers.
The backbone multicast video bandwidth is shared (unlike VOD’s unicast video bandwidth), but a home with 2-3 HD PVR set top boxes can draw 80-120mb/sec of the 100mb/sec. they have from the backbone to the home and have little to nothing left for voice or data.
Both Utopia and iProvo have already considered moving to MPEG4 to avoid the HDTV problem.
Much like a FTTH PON system, cable companies can continue to split nodes into smaller and smaller segments (I would bet most Comcast nodes today are 500 homes or less.) Bring an additional strand of glass on line (or pair) and you have 250 homes sharing, bring another and you are down to 125 homes (with no change at the home or in the coax from node to home.
I’ve said many times, fiber rich networks should NOT count of more bandwidth or fiber speeds to save them from cable companies.
For every step ahead they are for having fiber to the home they are an equal number of steps behind because all new services are developed for existing networks. Just like Apple…software is developed for PC’s and then ported to Apple later. Applications are being developed for cable not fiber. Things like CableCards, Tru2Way, and a host of interactive applications will all hit cable first.
Cable had HDTV and PVR and HDTV-PVR’s first. Developers would rather develop for a 60 million home market before the 1 million home FTTH market?
Don’t count cable out. It’s got lots of life. When it does peter out, they will have migrated to fiber if they needed too.
The problem with node splitting is that it can only do so much. Cox is down to 150-200 on a node and still can’t deliver LAN-like speeds. The problem is the coax. The medium can only support 4Gbps per segment. 2Gbps of that is easily eaten by video presuming you’re using SDV and DTA boxes. It’s probably more like 3Gbps. Leaving just 1Gbps to serve 200 customers, all of whom want 50Mbps Internet service? It would take just 1/10 of those customers being online to saturate the last mile presuming no voice calls (at 1Mbps a pop).
Fiber, on the other hand, has 10Gbps last mile available right now with 100Gbps right around the corner. They don’t have to compress signals into a muddy stream of pixelation (seriously, check out Comedy Central on Comcast sometime) because they don’t worry about bandwidth. Owners of HD sets are going to demand better picture quality and MPEG-4, while much better than MPEG-2, can only do so much.
Bear in mind that fiber installs recently outpaced cable. Product developers know that too. They also know that the transport layers could care less what medium they travel over.
In any network their is going to be bandwidth sharing at some point. In Utopia’s case they have 5.2Tb of switching fabric(this number taken from original claims made when they where pitching the network).
Its actually very easy to bring more bandwidth to the to any particular CO box. The trick is getting the bandwidth past that point. This is one reason why Utopia’s fiber is so flexible service options wises a business that needs a 1Gb dedicated Lan connection can be wired into the utopia setup very cheaply without effected anyone else on the network.
FTTN is insanely distance limited, 1000yards for full speed past that it goes down quickly. In order for Qwest to install FTTN everywhere they would have to install 6 times as many CO box’s as they have now. FTTN will be used in green fields and any place where it is politically convenient to prevent things like utopia.
Docsis 3.0 will not be fully used by the cable company’s as its a trade off for them, every time they bump the network speed on it they lose 3 HD channels or 9? SD channels. Satellite will be pushing over 100 HD channels very shortly Comcast and friends can not afford to take HD channels off their system and remain competitive with satellite. Comcast is already overcompressing their HD content in the vain effort to increase the number of HD channels they can fit into a single QAM 256 stream. If their is anyone that needs new direction its the cable industry, their bandwidth crunch is very real and its very here and now.
I had an intelligent comment to offer until I read the first commenter’s assertion that you (YOU!) have no idea how the oh so complex world of “The Cable Network” works. After reading that, I could only laugh long and hard.
Obviously not a longtime reader of the blog.
Also, I’m a little surprised to learn that Comcast is only now getting to the switch off of MPEG-2. I assumed it had already become the standard for larger companies.
jason: The move to MPEG-4 in video services is relatively new for all video providers. While many of us have used it extensively on our PCs (DivX and XviD are the more common implementations), it takes a lot of money to upgrade the transmission facilities to do it natively.
I guess for purposes of completeness to my previous comments, the FTTN system qwest would likely be deploying is ADSL2+ or otherwise know as ITU G.992.5.
and DOCSIS 3.0 info
and as to the over compressed comment
shows exactly what comcast is doing in order to create more room for docsis 3.0 and more HD channels. comcast needs to move to mpeg 4 quickly and do some serious updates to the last mile of their network.
Very interesting links.
Two points of clarification. I just want to restate luminous comment about the IPTV Multicast. This tech would be equivalent to the SDV on a coax system.
Also there was mention of 550 Mhz systems. I would think most areas around here that were ATT rebuilds are better than that. I know one of the reasons ATT first did their tests here was because of TCI’s better setup.
Besides that in my area on of the orginial test areas for ATT@Home (North Ogden area as well as SLC) they completely rebuilt the system in my neighborhood. They knew they were going to offer phone and data over the system. They came through the neighborhood and ran new conduit and coax to every home.
Now I don’t know what happened to locations after Comcast’s take over of the system or those systems that were not TCI in the area which Comcast took over later.
There is a way to check if your system if 800Mhz or 550Mhz still. You can look for the franchise ID on your bill. Then go to an FCC website and look it up. It will tell you what kind of system you are on. I don’t remember the link to the web site at the moment though.
I am not saying that Coax is the end all just noting that we are pretty lucky in the Wasatch front to have a very modern HFC system. Comcast has been a very aggressive about upgrades and keeping bandwidth on par with competition. So I don’t have a big problem with them on that front. They have also played pretty nice at all of the Utopia meetings. My problem is with other companies that seem to sit on their hands until they absolutely need to move and have no forward vision. Mostly because their hands are tied to the chair because of bad mismanagement over the years.
For those that remember ATT had said after the HFC network was built out they were going to allow 3rd party ISP’s to offer service. That dream died when Comcast did a hostile take over of the ATT Broadband side of things. I found it interesting that they then jumped on the Utopia build out only a few years later. Then left around the time of the SBC merger and other thing.
Good story of Qwest calling their connection “Qwest Fiber Optic Internet Service”
Copper is not aging. If they used 100% of the bandwidth on one copper cable for data instead of having it split between video and data, we’d have hundreds of megs per second in bandwidth.
That’s a pretty big ‘if’ given what is consumed by analog signals alone.
Remember that Comcast (and all cable operators) are moving quickly away from analog and will deliver all video as digital.
Beyond that, cable will be moving to switched digital video in the future. Thus sending only the watched channels as an IP network does today.
Both of these are proven cable technologies available to cable operators (at an upgrade cost) when they feel they need it.
Cable can deliver more bandwidth today (Comcast is launching a 50mb/sec. package in about a dozen cities this year.)
That’s faster than any UTOPIA provider (except Mstar)?
Xmission offers a 15Meg and 50Meg home service.
Business is 30/50/100. I also believe if you want more they can work with you.
all of the utopia options are sync speeds so the upload is as good as the download where as the comcast 50meg down only has a few meg upload making it unless for many neat 2 way services, video phones, game servers, file sharing etc. Don’t get me wrong tho I understand that without replacing their network they are stuck at that, and that docsis 3.0 is still a huge upgrade for people stuck with them. I still don’t buy that the cable system will be able to provide the needed amount of bandwidth in the coming years. once you add together all the different things people will demand in the coming years cable just can’t cut it.
I guess I’d agree with the main premise…fiber has more bandwidth than coax.
I’m not sure when the time will come when coax will not have “enough” bandwidth. I’m fairly sure it is not any time soon.
With fiber to the nodes and the ability to continue to split nodes to serve fewer and fewer customers I believe coax has decades of use left, even as individual bandwidth usage increases.
There is no technology issue with changing the up/down ratio of coax. With a piece of coax being able to carry over 3 gb/sec. it’s hard to see them running sort any time soon if they just continue to split nodes to server fewer and fewer subs per coax section.
While the service providers say the speed is synced, I’ve run hundreds of speed tests and never get anything near a sync speed? Does anyone else feel they are getting 15mb or 20mb in each direction?
I don’t think cable can node-split as much as they need to. Cox is leading the field at 150-200 per node whereas other providers run anywhere from 500 to 2000 per node. You already see a single triple-play household eating up a full 100Mbps connection on systems like iProvo and UTOPIA and that’s a point-to-point network. Sure, they can free up a lot with DTA boxes, SDV and MPEG-4, but by the time it’s deployed, they have to offer even faster speeds to stay competitive and all gains are erased.
I think the problems with hitting full sync speed has to do more with the speed test servers themselves than anything else. They’re tuned for much slower connections at asynchronous speeds. Many web servers aren’t geared towards those kinds of speeds either. The best way to truly saturate 50Mbps+ is with multiple data streams at a time.
In reference to your question about seeing the advertised speeds. I don’t currently subscribe to a Utopia service (I would if I could). So I can’t say what kind of speeds happen from local providers but most speed tests have issues with higher speeds. Hence Comcast’s new speedtest.comcast.net which has different tiers for testing.
On another note the best test for XM customers would be to download something from mirrors.xmission.com. This may even be an ok test for those not on XM’s network.
For reference my place of employment has a 1GigE link to the Internet. My desktop is connected at 1GigE to our backbone.
I just did a test to mirrors.xmission.com and for the most part it sustained 2.6MB/sec. A test to mirror.cs.utah.edu downloading an OpenSUSE ISO clocked in at around 4.4 to 6MB/sec. (Disclaimer on the utah.edu test. That isn’t traversing the Internet just the UEN network.)
While downloads from various places usually are pretty good they don’t usually go faster than 1 Megabyte a second and thats only on the really good mirrors. At that speed even a 10Meg connection would work.
The thing for me is when I work at home on my Comcast connection and VPN to work. Because of the asymmetrical nature certain tasks are slow. If I were on a 15/15 connection I would be a very happy camper.
Please don’t get me wrong…I’m a happy camper on iProvo/Broadweave. My download is typically close to 15-20mb/sec. on speed test sites and the upload speed is very fast and nothing I would complain about. I’ve just rarely seen it test at close to the stated rate. It’s still plenty fast, no complaints here.
There is little to stop cable companies from splitting nodes as small as needed. That is the operative word here…”as needed”. Cable company nodes are the size they are mostly because the company has made a decision that they do not “need” to split more or provide faster speeds to get/keep customers.
Should customers leave en mass due to slow speeds, they would increase the speeds.
There is lots of life left in coax systems.
how many times can they afford to split a node before the costs become silly?, Can we be sure that if a node needs splitting that it will be split. Cable rates are already out of hand, who wants to have the costs of their cable bill increase when the cable co uses these node splits as an excuse to increase their cable rates. With competition from satellite how far will the cable co split nodes and stay competitive price wise? How well does even a 200 customer node handle peak usage?
Now lets not compare to the best in the industry lets compare to the worst in the industry Comcast!! does anyone really believe with a straight face that Comcast with ever split their nodes to that extent? The only reason their doing docsis 3.0 is because it is cheap to implement.
I believe Comcast (and other companies) will only spend what is required, when it’s required.
So as long as they have all the customers, they will not need to split nodes. But you should rest assured that if it means keeping their business, their survival, they will spend what is required to keep up as best they can. Splitting a node is not that difficult or expensive, it’s just not considered unless needed. Subscribers wanting faster service is not “needed”, subscribers disconnecting because of your slow speed…that’s when it’s needed.
While Fiber is better and cheaper than Comcast today….it’s VERY IMPORTANT to remember that Comcast has almost all the subscribers.
So IF it were about speed, UTOPIA and iProvo would not have 20% or less of the customers.
The FASTEST speed means very little. You must provide a speed that the vast majority of customers feel is “fast enough”. That’s what Comcast and others do very well.
If Comcast customers were unhappy with their service (for any reason), UTOPIA and iProvo would have the bulk of the subs in wired areas….and they dont!!!
I suspect the answer to your question, “How does a 200 subscriber node handle peak usage?” is “Good Enough!”, If that were not true, customers would be lining up to sign up for the faster, cheaper UTOPIA service.
You can talk about Comcast having bad service or problems, but if customers have somewhere else to go (UTOPIA in some areas), and they are NOT going….that say’s you must be wrong!
Think about it!
“So as long as they have all the customers”
i would like to know who is going to take those customers… Qwest sure isn’t their FTTN system is a joke that is going no where, their are no other cable companies that overlap into their territories, and Utopia while great is only in a short number of city’s. Meaning that for the most part the status quo is good enough for um. if a cable customer doesn’t like his speed, uptime, customer server he most likely is screwed but pretending he is one of the lucky people who can get DSL he can choose to receive similar crappy customer service, slightly improved uptimes and at best 1/10th the speed of his cable connection. Thank god for competition.
As for Utopia no one knows about it, the only advertising they have had is a few city meetings and a couple short mentions on a few isp’s web pages. While Qwest has been heavily advertising services you can’t even buy (ADSL2+), and Comcast has probably spend as much on marketing in the Wasatch front market sense Utopia started as Utopia has debt.
I was only considering areas where UTOPIA/iProvo were available. The point being EVEN IN THOSE AREAS, Comcast has the overwhelming majority of the customer.
Service providers have done heavy door to door in many areas. UTOPIA’s failure to market successfully (or that of their service providers) is no one’s fault than their own.
Perhaps a leadership team with cable/telephone experience would not have let that happen?
I predict that you will see little change. UTOPIA’s penetration will remain below 30%.
If anything this supports my saying in other posts (see my comments on the Forum under “Why UTOPIA should NOT be building (yet).”) that UTOPIA should NOT be building any more network until they figure out how to sell the services. By providing customers a service they want at a price they will pay. If UTOPIA can not sell to the tens of thousands of homes they are already built past and able to serve, they will do no better by building more plant.
UTOPIA (and it’s service providers) needs to demonstrate that they have the skill set and ability to market a product people want, at a price they are willing to pay. They can do that in existing footprints. Adding customers in already built areas is much cheaper than building more plant.
…and so far, I think they have NOT done any new construction. Just finished up areas already mostly done. But I have not seem a MARKETING push in areas like Orem, Payson & other built areas. I guess they are waiting on the “magic bus” to help with that?
But if they CANNOT sell at the rate needed to be successful, they should (and likely will) sell the network to someone who can.
Comcast offers free DTAs to certain customers
This is part of the plan to reclaim bandwidth from video (by going all digital) giving them more bandwidth for data services and HDTV.
More evidence (for those that need it) that you should not consider coax as bandwidth limited:
News story on Insight Cable testing a 100Mb download:
“Insight Communications is testing a download service in Louisville that is five times faster than anything currently available to consumers.
Spokeswoman Sandy Colony said the company is testing the 100 Mbps service at a physician’s office, but she declined to say when the service might be rolled out to other customers.
The company’s top download speeds are now 20 Mbps. The fastest speeds nationally are generally considered to be from fiber connections and are typically about 50 Mbps to residential customers.”
I believe fiber is better than coax, but we should not count coax as out. There is plenty of life left in coax. Not as good as fiber, but good enough to meet the need for years to come in my opinion.
Fiber Snobs: Read it and Weep!
Yes, we deserve a small slice of humble pie. You will note, however, that the networks are still very heavily encumbered by the video content they deliver and some design flaws limit upstream speeds.