Business Week article about how Verizon wants to reserve 80% of its network bandwidth for itself, caused a minor ruckus yesterday as one after another, everyone brought up the issue of network neutrality and started the finger pointing. Given the Bells past record, that was hardly a surprise.
Documents filed with the Federal Communications Commission show that Verizon Communications (VZ ) is setting aside a wide lane on its fiber-optic network for delivering its own television service. According to Marvin Sirbu, an engineering professor at Carnegie Mellon University who examined the documents, more than 80% of Verizon’s current capacity is earmarked for carrying its service, while all other traffic jostles in the remainder.
It did not make much sense, given how Verizon is building its network. They are using passive optical network technology (PON), where the transport bandwidth of a fiber is shared amongst N-homes, where N-varies between 8-and-16 homes. Verizon FTTH is split into two parts – one part for video transmissions and the other for data-Internet traffic. The video part of has spectrum limitations and that is why company can currently offer only upto 120 channels. The video signals are sent to your TV set pretty much like how cable companies send their signals to consumer homes. This is TV, not IPTV, and it sucks up most of the available capacity on the network.
IP Democracy points out that the video part of the network will be 3.5 Gbps capacity whole date will have 622 Mbps. Data (Internet) part of the network needs less spectrum, but PON technology for now limits the amount of bandwidth that can delivered per household. In theory, a 622 MB/s link that is shared among say 32 households, means each home gets about 20 megabits/second. Business Week story wasn’t all that clear about the crucial difference between the “TV” and the “Internet part.”
Still, not sure about this, I decided to check with a bunch of smart folks who know more about networks and the Internet than I do. Vint Cert, who is now Chief Internet Evangelist at Google (in an email) said, “My understanding is also that VZ wants to reserve BW for video to compete, presumably, with cable companies.” Going forward, however, the architecture adopted by VZ is going to become a bottleneck, says Cerf (in an email.) “The consequence of this is that the “broadband” access to Internet will be limited in capacity and likely not symmetric, unlike plans of the Japanese NTT-east to offer 100 Mb/s symmetric capacity access to Internet, for instance. ”
Dewayne Hendricks who is quite wise in the ways of the Internet wrote back in response to my query…. “It seems to me that one of the unstated things in this controversy is the meaning of ‘broadband’. Take a look at this blog entry and you’ll see what I mean.” If you go buy into the definition of the incumbents, then 20 megabits/second is plenty. “If like some of us, you see broadband being something like 100 Mbps symmetric and above, then we’ve got some real problems with their current and future offerings and network neutrality takes on a entirely different meaning,” says Hendricks.
Cynthia says it well when she writes, “I’m not being an apologist here for Verizon or the telcos — far from it. But, if you’re engaging in a political fight, at least be accurate with your aim and don’t drum up problems that don’t exist.” My sentiments exactly!
I think the most amusing part of this whole drama has been overlooked by all. So lets assume, Verizon starts selling you video-on-demand movies for $3.99 a movie over its analog TV pipe. Yet at the same time, using the Internet pipe, I can download the same movie (or even stream it) from an online service like Vongo or Movielink for $2.99 a movie. Wouldn’t VZ end up competing with itself. Maybe that’s why the incumbents want the network “tolled.” Just a thought….
IPTV is almost certainly why the telco’s want two tiered internet. Without out two tier internet any telco’s IPTV service will be uneconomical against broadband TV.
As you correctly note, FiOS’s video service is not being delivered to the home over the internet pipe, rather, it’s being distributed at a different wavelength (1550 nm) over the fiber from the ‘head end’ to the customer’s NID where it is converted to traditional coax. The data goes over 1310 & 1490 nm.
Some good details on how this works can be found here:
http://www.exfo.com/en/support/WaveReview/2004-November/WRarticle6.asp
http://en.wikipedia.org/wiki/Fios
http://www.dslreports.com/forum/remark,15277912
http://telephonyonline.com/home/news/telecomverizonplanned_video/
I believe that Dr. Sirbu’s numbers were based on assigning a speed value (of 3.5 Gbps, according to http://arstechnica.com/news.ars/post/20060202-6104.html) for the video services, though I’m not sure how he came up with such a high number.
Frank
OM said ” I think the most amusing part of this whole drama has been overlooked by all. So lets assume, Verizon starts selling you video-on-demand movies for $3.99 a movie over its analog TV pipe. Yet at the same time, using the Internet pipe, I can download the same movie (or even stream it) from an online service like Vongo or Movielink for $2.99 a movie. Wouldn’t VZ end up competing with itself. Maybe that’s why the incumbents want the network “tolled.” Just a thought….”
I don’t agree entirely. You can get the same content at a cheaper price but the experience is different – one on a big TV and Vongo’s on the PC. Different audiences as well – you’re assuming that the audience will adopt Vongo en masse like television (not likely).
Not knowing specific Verizon plans, let me share my views of where the PON (FiOS) technology is going to and where video transmisison is moving to over PONs.
Yes, Verizon had analog video over the third band for their BPON deployments and it has/had its limitations as discussed. However the shift is now to deploy GPON. GPON will be 2.4Gbps downstream and 1.2Gbps up. There is not expected to be any third wavelength video, although the standard does allow this. Calculating 64 splits (end users) per GPON you get a raw data rate of 37.5Mbps averaged per user.
Now, as video will not have a separate stream, it will then be IPTV. IPTV for HD is running somewhere between 8 and 13Mbps/channel. The goal is 5-8Mbps but let’s expect a realistic 8. Standard digital video streams are 2-5Mbps depending on quality, etc. As Telcos must compete with a “better than cable service,” and they are new to this arena, expect them to default to the higher bitrates/quality. As the Single Family Home (Unit) or SFU in US has ~5 TVs and each will need to have a set top box. As HD is expected to be on 1-2 TVs in an SFU (3 is a better and more conservative planning number), and all TVs may be on at any given time, the bandwidth need is between 25 and 31+Mbps. So, a Verizon claim of needing a vast amount of bandwidth (BW) for video (their) service is somewhat valid.
Now, looking at Ethernet and IP and the concern for QoS, then a Telco will want to ensure that their video delivery, which once again must compete with Cable and be at least as good quality and delivery (uninterrupted) wise, then you have a need for BW management and reservation.
They Telco argument does leave factors out… I believe the rule of channel watching is also an 80/20 rule. That at any given point in time 80% of the viewers are watching the 20 most popular chanels. In fact, there is astatistic out on the web that shows given 5 TV sets in a home being watched simultaneously that any 2 of them may be on the same channel. So, using IGMP and the ability of Ethernet to broadcast the same packets to multiple targets, there is the chance to save BW on the PON link to one home. Given that the BW will be managed into the ONT (PON head end, so to speak) then the IPTV data being delivered to be then distributed to the 64 user homes will not require that there be a full video feed of ~30Mbps per home for ~1.9Gbps. However, this is statistical surmising, and something a planner canot bet their QoS on, so best to presume all BW is being used independent of other BW. A bit of overkill, but something of a valid conservative approach for a Telco coming in as a disruptive entrant versus a Cableco.
So, wrap up: Verizon will want to protect this high amount of throughput for their services and QoS needs. The remianing BW will be open to other services and management overhead. I sense that the Verizon concern is two-fold in that 1) they do not want the IPTV BW from OLT to home to be compromised. Then 2) looking from the OLT into the Metro space (where BW is restricted and the infrastructure is in need of build outs) there is a need to manage and control the BW usage as well. This is where I see the Telcos wanting to charge money for high BW demanding feeds. Is a toll for using their right of way. Is this valid? Time will tell.
However, and I always wondered about this, why the Internet and BW was always thought to be “free” once you had a pipe to the Service Provider. I have wondered why someone did not come up with a much lower rate link (DSL) cost and then charge for the number of Mbits sent per 24 hour period and then a surcharge for peak (great than x number of Mbps) usage. Means that a user, with a desire to watch a streaming Yahoo video for $2.99 may pay a surcharge of $0.xx to cover the higher rate required for such a higher QoS connection and then may pay a slightly higher monthly bill for the total Mb used over the month. So, who pays? The consumer. Not the Service Provider. To have them pay would be to reverse the view to also say that they are charged for Mb usage (albeit at per a very different billing schedule) and peak demand. This may happen to then reduce some of the ocnsumer impact. But the positive impact, if consumers are charged, is that the physical link (DSL) ought to be much less per month for the service provision and connection fee.
This is in line with old time Telco billing and seems to me to be more in line with Network usage, which is where the true costs will be over time… making the pipes big enough to get/send the data you want at the rate you want. I wonder if the Europeans, with their focus on higher costs to consumers, may take an appraoch such as thie for VDSL and PONs.
Om, I’m glad to see your debunking here.
BTW, the 622Mbps is (or could be) shared dynamically, so that’s ~20Mbps/customer worst case. Under normal circumstances I think 100Mbps/customer would be possible.
Wes,
PON is a static split, so the max bandwidth per user is fixed by the split. So, if they split 64 then 37.5 is the max.
This will not be enough BW for long term video delivery, so they will need to reduce the split.
This static spliting is the reason I don’t like PON, but prefer active networks, where the bandwidth is truely shared.