Earlier this evening, a Verizon spokesperson sent out a link to a post on Verizon’s blog about the recently announced 100 Mbps broadband offer from Cablevision, a Bethpage, N.Y.-based cable company. The offer was a “parlor trick,” Verizon said. That argument is one we’ve heard often […]

Earlier this evening, a Verizon spokesperson sent out a link to a post on Verizon’s blog about the recently announced 100 Mbps broadband offer from Cablevision, a Bethpage, N.Y.-based cable company. The offer was a “parlor trick,” Verizon said. That argument is one we’ve heard often when it comes to cable broadband.

With today’s technology, you don’t have to break much of a sweat to deliver 100 Mbps to a few customers. But given the inherent limits of the cable platform, a cluster of bandwidth junkies living near each other could be a real problem. One estimate is that a single 101 Mbps customer would use some 60% of the capacity in a neighborhood. Other users? Outta luck.

DSL Reports’ Karl Bode has the best response to that charge. “We watched Cablevision completely take the brakes off of DOCSIS 2.0 by offering consumers uncapped (in the true sense of the word) 30Mbps connections, and the company managed to survive the last three years without any sort of ‘coaxapocalypse,'” he writes.

I had the same thought and had a back-and-forth with Verizon’s spokesperson, who essentially said that GPON is a better technology option and they deliver what they promise. Fair enough — I have yet to hear anyone complain about bandwidth delivered by Verizon’s FiOS. At the same time, no major hue and cry has broken out over Cablevision’s cable broadband performance. Cablevision is selling broadband to half the homes in its service area, so any problems would be aired pretty quickly.

To me, Verizon’s response to the Cablevision offer shows that they are finding themselves on the back foot. They don’t have a 100 Mbps service. Cablevision’s offer puts pricing pressure on Verizon’s 50 Mbps offer, which costs upwards of $140 a month. Any price competition gives Verizon, which has spent $23 billion on its FiOS buildout, fiscal heartburn.

I bet we haven’t heard the last of this battle of broadband service providers. That’s OK — I’m glad to see that at least in one part of the country, broadband competition is working and resulting in better options for consumers. I wish that were the case in San Francisco.

You’re subscribed! If you like, you can update your settings

  1. “I wish that was the case in San Francisco.”
    Or anywhere in the north bay for northern cal for that matter. Why the frigging heart and edges of silicon frigging valley is allays on the absolute ass end off of every roll out schedule is pretty dam irritating.

  2. Ok I admit it, it’s the 3rd time this month SBC has called to offer me, no really, there new Super Fast DSL, instead of the dam fiber I want (and tell them not to call back without)… and of course comcast is nowhere near to doing anything exciting in my neck of the woods. I ‘Hate’ the lack of competitive pressure in these region.

  3. You can’t analyze a technical issue by surveying blogs for opinion?

    Cable bandwidth is inherently limited because the local loop ‘cable’ is shared! Cable was not designed for internet data, it was installed to deliver a fixed (and limited) number of one way TV channels. Get a couple of customers in the same area running bit torrent and all other users will slow to a crawl.

    The only way to deal with the technical limits of cable is to throttle users with caps… which totally defeats the point of having 100 Mbps speed. It just means you are able to hit the cap within a couple of hours instead of taking the best part of a month to hit it.

    Any comparison between cable internet and ANY competing technology that uses a dedicated channel between the exchange to each customer (ADSL) will always see cable as a loser! Obviously if you’re a cable company it’s better to perpetuate the illusion that isn’t the case.

    1. Every architecture has shared infrastructure somewhere. ADSL2 can give you 24 Mb/s dedicated to a single customer – but the uplink is shared. Put out a DSLAM serving 512 subscribers with a single GigE uplink, and you’re averaging 2 Mb/s per customer. GPON shares 2.4 Gb/s among 32 or 64 subs – so you’re averaging either 75 or 37.5 Mb/s per sub – but again, chances are you’ve got a single GigE uplink, averaging 31.25 or 16.625 Mb/s per sub.

      Everyone shares, and everyone has oversubscription ratios somewhere. “Dedicated” versus “shared” is a red herring – the questions are what is the overall capacity of the system (end to end), how is it engineered, and how competent is the network operator in managing the network capacity (as Josh Auerbach alludes to below). And how cost effective it is to deliver that capacity, as Luis Rodrigo points out below.

      1. Actually, the uplink capability on Verizon’s GPON platform far exceeds the capabilities of a DSLAM architecture. Their latest deployments use Motorola Optical Line Terminals. The Motorola uplinks scale to 20 Gbps with a roadmap to 40 Gbps. When you consider the burst capability per PON and the substantial uplink headroom, cable doesnt stand a chance in the long run. The key word is “long run”. Cable operators can bond 2 to 4 cable paths to achieve significant throughput but they cant possibly scale it for sustained general service offerings. Not a problem for them today, but it wont be long before the fat lady sings for cable.

      2. PBXTech – Yes, GPON uplink far exceeds DSLAM uplink, but again, the limiting factor is engineering, not theoretical capacity. While a GPON OLT is theoretically non-blocking all the way through (switching capacity and uplink capacity equal to or greater than access capacity), I would wager that VZ isn’t dropping 20x10GbE links onto every OLT.

        In the long run, we are all dead… In the slightly less long run, there’s nothing that limits DOCSIS bonding to only four channels. DOCSIS 3.0 specifies 8-channel bonding, and if memory serves, TI and Broadcom have both announced silicon supporting this, available to ship 3Q09, upping the total data bandwidth on DOCSIS 3.0 networks to 300 Mb/s. Subsequent standards can increase the number of bonded channels as necessary.

        Cable HFC plant with DOCSIS 3.0 and subsequent standards, combined with total digitization and switched digital video to free up QAMs and some aggressive node splitting will provide pretty darn good capacities. No, it won’t match GPON or future PONs – physics is physics, and fiber has more raw capacity than coax – but it’ll be good enough, significantly cheaper, and require gradual capex spending rather than throwing $B at the network in a small number of years. An 8-channel bonded DOCSIS plant with 20:1 oversubscription, 125 HH per node, and 50% take rate gives average capacity of 93 Mb/s per user – on par with GPON with a 32-customer split. And the MSO will have a number of tools to increase the capacity – higher degree bonding, node splits, adjustment of oversubscription ratios, load balancing across colocated nodes, some of which require no capex.

    2. “Cable bandwidth is inherently limited because the local loop ‘cable’ is shared! Cable was not designed for internet data, it was installed to deliver a fixed (and limited) number of one way TV channels. Get a couple of customers in the same area running bit torrent and all other users will slow to a crawl.”

      you dont know anything about cable.Cable was FIRST and the fastest internet service since 1997? and it will keep it like that.now we moving to docsis 3.0.in another 5 or 6 years we will be with docsis 4.0 and u`ll see a huge improvement Cables is FAST,SIMPLE and is EVERYWHERE!

  4. it makes you wonder if they are simply saying that because they have competing services or if it is actually true. I would hope the speed claims are correct but seems like such a jump from the average service in the US that it seems unlikely to be actually true. I have a feeling that this is a slight over exaggeratoon of the typical steady transfer rates. I feel lucky to be able to get a little over 10 Mbps. I would just about kill for 50 or 100 Mbps! You hwarthat , acid

  5. Oops, the last few “words” in my previous post are typos. Sorry, I’m on my iPhone.

  6. While Silicon Valley isn’t seeing 100 Mbps service from Comcast or AT&T yet – Comcast is now turning up their improved Cable Modem services in several cities here in Silicon Valley.

    After resetting the Cable Modem at home today, I noticed that my existing “Blast” service which had been topping out at 5 Mpbs downstream last week has indeed doubled to rates of 11 Mbps to 14 Mbps downstream and 3.5 Mbps upstream.

    There is an option to move to a Docsis 3.0 modem for $10/month more for speeds of 22 Mbps/5 Mbps. Haven’t been tempted to try that yet.

    So while I wouldn’t declare victory yet, we are seeing improvement. An encouraging sign indeed.

  7. Benoit Felten Thursday, April 30, 2009

    One could argue that Verizon calling out on Comcast is the pot calling the kettle black. GPON has a downlink of 2.5Gbps shared between (most likely in the case of Fios) 64 customers. Do the math, you won’t deliver 100Mbps to all customers either.

    Now I don’t know exactly which implementation of Docsis 3 cablevision has and how it shares the bandwidth, but it’s most likely not all that complicated.

    As for price wars, it gives Verizon a heartburn for the wrong reasons. I’ve been doing a lot of analysis on the FTTH business plan and in fact am publishing a report on that topic next month, and I think Verizon has a number of options to optimise their business model even if there is competitive pressure…

    I’ll be happy to share that with you Om if you’re interested!

    1. Vlad (Small Business Blog) Benoit Felten Thursday, April 30, 2009

      Verizon lacks (so far) the so-called “geek option”, the one that has made the Boost attractive to me (and, probably, other geeks). With all the nice speed and all – Verizon doesn’t let you run OpenDNS and anything on ports 21 and 80. From my personal small-business-owner perspective – they may have 50Gbps connection – it’s just as useless as 50MBps.

      However, I am myself very much interested in your report. Hope Om will publish a link when it’s ready.

      1. Concerning FiOS, it is true: port 80 is blocked. However, port 21 is OPEN on my end; i know this as i run an ftp server. I agree, the bandwidth offered isn’t everything, but what you can do. at a time, the tos stated that you could not abuse fios for high-volume purposes, but nothing about running servers. Except it may be different now. Nevertheless, verizon has been known for liberal use, and most celebrate verizon for taking a stand against the RIAA when they subpoenad customer info. <- although that part may only apply to potential illegal downloaders.

  8. Josh Auerbach Thursday, April 30, 2009

    I’d echo Benoit’s and others’ comments. Cablevision now clearly has the technical capacity to deliver 101 Mbps in the last mile. Whether or not they take the required steps (in terms of node-splitting, backhaul from the head end, etc.) to provision that bandwidth adequately is a separable question, but certainly not a question on which Verizon would have much insight now.
    The real test will be comparing Cablevision’s speed — as delivered — to Verizon’s. My guess is that Cablevision will do just fine in that comparison. It’s not a “parlor trick” if it works.

  9. I live in Brooklyn, NY (An area served by both Cablevision and FIOS). I have had Cablevision service for 3 years now. I have been subscribed to the plain 15/2 plan and have consistently received those speeds. The network is rock solid. I consider myself a heavy user and get downloads of 1.5 MBps consistently. One of my neighbors was sold into FIOS with the “faster speeds” argument. He reported that his connection was good in the beginning, but now after many other people got it, it got slow.

    1. If a FiOS customer’s bandwith gets “slow” then they need only call and report the trouble. PON systems will absolutely ensure a committed rate to each subscriber. It’s how they’re designed.

  10. While many of these discussions tend to focus on technical superiority of one architecture over another, what I see missing is the discussion about the economic attractiveness of the architecture. It seems to me while Verizon’s architecture arguably might be superior, if looked at from a financial ROI or IRR perspective, I would think that the cable architecture is much more attractive. Hence, there is more leeway to offer a compelling price point to the subscriber.

Comments have been disabled for this post