1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

SDV FAQ

Discussion in 'TiVo Series3 HDTV DVRs' started by bdraw, Jul 3, 2007.

  1. Dec 21, 2007 #841 of 2401
    bicker

    bicker bUU

    10,382
    43
    Nov 9, 2003
    Georgia
    Of course it is. Look: You and I disagree. You can type in your drivel all day long and it isn't going to get any more convincing. I think you're wrong. Get over it. You've had your say; I've had mine. Time to move on.
     
  2. Dec 21, 2007 #842 of 2401
    mikeyts

    mikeyts Stream Warrior

    2,389
    2
    Jul 10, 2004
    San Diego,...
    What either of you thinks is irrelevant. The cable providers and the companies developing SDV system have done research and they're convinced. Real, hard data acquisition and analysis--no armchair BS. They pumping many tens of millions of dollars into this and imposing it on us all. It's a gamble, but their only alternative is pumping many billions of dollars into running FTTH to a hundred million households.
     
  3. Dec 21, 2007 #843 of 2401
    bicker

    bicker bUU

    10,382
    43
    Nov 9, 2003
    Georgia
    Abso-friggen-lutely.
     
  4. Dec 21, 2007 #844 of 2401
    dswallow

    dswallow Save the ModeratŠ¾r TCF Club

    50,973
    31
    Dec 3, 2000
    Long...
    Except that there'd be sufficient additional outlet charges involved that the cable company would certainly be getting compensated for it. And considering just how many it'd require there may be an upper limit on the number of additional outlets they even allow for one account.
     
  5. Dec 21, 2007 #845 of 2401
    morac

    morac Cat God

    8,936
    19
    Mar 14, 2003
    NJ
    It can't be all that expensive if Joe Consumer can get a HD camcorder for less than $1000. Yes I know the industry uses more expensive cameras, but even if you're talking a $30,000 camera that should still be affordable to most content providers. As for encoding HD, my desktop PC can do that (albeit slowly) so I don't think that's an issue either.

    By the way the current list of HD channels is impressive and it's only going to get bigger. How cable companies will be able to provide all these channels is the big question. I personally think that SDV is not the ultimate solution and that the only way this can be done is by switching to MPEG-4 encoding.
     
  6. Dec 21, 2007 #846 of 2401
    lrhorer

    lrhorer New Member

    6,922
    0
    Aug 31, 2003
    San...
    That's correct, and I was a fairly significant contributor to that effort until a while back. As a CATV engineer helping to implement the Pegasus system and SDV I had to understand how SDV works and what the implications were for useage, and I still have some inside views on what's going on with the CATV company. The point is, however, they aren't spending a dime worrying about TiVo Suggestions because there isn't anything to worry about. The simple fact - not an opinion but a cold, hard fact - is TiVo suggestions do not add any significant load to the SDV system as a whole. They add nothing at all to the server farm's peak load, and additions to off-peak loads are not relevant in any case. They do potentially add network congestion issues to the local node, but once again this is hardest felt during peak load hours and the additional load is as well represented by any DVR, including the ones provided by the CATV system, not just a TiVo with suggestions.
     
  7. Dec 21, 2007 #847 of 2401
    lrhorer

    lrhorer New Member

    6,922
    0
    Aug 31, 2003
    San...
    I have news for you. $30,000 is cheap for an SD commercial camera, let alone an HD. The big networks don't blink an eye at six figures for a camera. The thing is, there aren't hundreds of national networks out there. Most of the content comes from small concerns, down to and including public access where Joe Bob and Cindy create a Merry Christmas video for distribution on the Public Access channel. The cameras the CATV company loans out to those folks cost less than $5000, and that's not liable to change.

    That's the point, and it is an issue. Anything close to "slowly" is not acceptable for a live broadcast. Of course movie channels and the like have the luxury of taking all the time they want, and they don't use a camera, at all.

    The computing horsepower to encode compressed HD video from a live uncompressed stream is phenomenal. Your computer just re-encodes already compressed video, which is much different. The last time I saw a benchmark (although this was several years ago), it took the fastest supercomputer in the world an hour to compress a 60 second commercial into a 1.544Mbps stream without visible artifacts. Of course now chipsets have been developed to handle the compression natively, and computing horsepower is growing at an exponential rate, but it still isn't quite dirt cheap.

    Well, yes, and no. The number of channels deliverable via SDV is virtualy without bound. In order to increase the number of channels, the CATV company merely needs to decrease the number of houses served by a node. The end limit for node expansion is reached when the number of receivers in every node is less than the number of channel slots on the node, at which point the number of deliverable channels is infinite as far as the distribution system is concerned. Therefore SDV alone is quite sufficient to deliver any number of channels one might wish. The wrinkle is MPEG-4 allows more channels in the same bandwidth, which means the node size doesn't have to be as small. This translates into more channels for less money for the CATV provider, so I have no doubt whatsoever they will be moving to MPEG-4 or some similar protocol in the not too distant future. It won't eliminate SDV, though. What's more, SDV allows for much more than just ungodly numbers of channels. It allows for essentially unlimited venues of interactive services from banking online to interactive classrooms with a live lecturer (rather than what passes for "interactive tutoring" today) to videoconferencing to voting online.
     
  8. Dec 21, 2007 #848 of 2401
    cableguy763

    cableguy763 New Member

    525
    0
    Oct 29, 2006
    Austin
    If a cable co goes to mpeg 4, they would have to have all new boxes. Pretty expensive option.
     
  9. Dec 21, 2007 #849 of 2401
    lrhorer

    lrhorer New Member

    6,922
    0
    Aug 31, 2003
    San...
    Yes, but I am a qualified expert in the field and you are not. Yet once again you resort to ad-hominem retorts rather than real, physical data.

    Let me set up a typical scenario:

    Moderately large city
    100,000 subscribers
    2.5 live viewers per household
    2 channel DVR per 10 households ignoring any TiVos whatsoever
    = 270,000 receivers
    500 scheduled channels (no CATV company yet comes close to this, even in the largest cities)

    Answer me this: How many of the 500 channels do you estimate are not being viewed by a single receiver peak viewing times? (I'll give you a hint: the answer is less than 1.)

    Changing out every single one of the 235,000 receivers to a 5000 channel super-DVR recording every available channel 24 hours a day on 10 times over won't increase the load on the server farm by any amount whatsoever, except to increease the number of broadcast addresses in the headers.
     
  10. Dec 21, 2007 #850 of 2401
    lrhorer

    lrhorer New Member

    6,922
    0
    Aug 31, 2003
    San...
    'Not if they are already MPEG-4 capable. I don't know the innards of the new boxes all that well, but even if they don't, a staged retrofit coupled with a staged rollout of MPEG-4 would be about the same as the current SDV rollout. At what point it becomes worth it is the question.
     
  11. Dec 21, 2007 #851 of 2401
    dswallow

    dswallow Save the ModeratŠ¾r TCF Club

    50,973
    31
    Dec 3, 2000
    Long...
    Not really. The SDV rollout doesn't involve replacing the cable boxes at every household. If those boxes can't handle MPEG-4, an MPEG-4 rollout would only become advantageous when all subscribers to any given channel had MPEG-4 capable receivers.
     
  12. Dec 21, 2007 #852 of 2401
    mikeyts

    mikeyts Stream Warrior

    2,389
    2
    Jul 10, 2004
    San Diego,...
    Interesting. BigBandNetworks have some papers online describing field trials of the technology which would seem to uphold their premise that it does work. There are papers here and here about trials run in 2002 and 2003. In their studies the numbers strongly support the premise that great efficiencies can be fairly easily acheived with SDV in the real world, which seems intuitively reasonable. Indications are that number of channels in demand does not grow very much after a certain point as you add subscribers. Given a fixed population of subscribers, growth of demand as you add channels is also flat after a certain point. I'm not sure against what population of subscribers, but they think that you can support a 500 channel system with 276 active streams (which they calculate to require 28 QAMs, but they're obviously talking about SD channels packed 10 or so to the QAM). With "maximum broadcast bit rate" HD programs, you'd need 138 QAMs for 500 HD channels. Of course, a 500 channel system put together today would probably have no more than about 100 HD channels with the rest being SD.

    BigBand did do these studies, but they're trying to sell the technology which no doubt colors their conclusions. What evidence do you cite to back your assertions?
     
  13. Dec 22, 2007 #853 of 2401
    mikeyts

    mikeyts Stream Warrior

    2,389
    2
    Jul 10, 2004
    San Diego,...
    TWC Austin rolled out SDV over a year ago. I wondered if they'd published anything about their experiences. There are a few tidbits in the following Multichannel News articles:Apparently, at the time of the second article, they were switching 175 SD channels and 8 HD ones on 8 QAMs (normally about enough bandwidth for 80 SD channels or 16 HD ones). Impressive.

    EDIT: I just noticed that the entired second article (which was published first) is included as a sidebar in the first piece.
     
  14. Dec 22, 2007 #854 of 2401
    ah30k

    ah30k Active Member

    2,211
    0
    Jan 8, 2006
    The fact that you don't understand the 'node' concept with SDV shows you are not much of a qualified expert.
     
  15. Dec 22, 2007 #855 of 2401
    mikeyts

    mikeyts Stream Warrior

    2,389
    2
    Jul 10, 2004
    San Diego,...
    I also noticed the following at Multichannel News:Pointed out there as here is the fact that use of it would require distribution of boxes capable of MPEG-4 decoding.
     
  16. Dec 23, 2007 #856 of 2401
    lrhorer

    lrhorer New Member

    6,922
    0
    Aug 31, 2003
    San...
    I understand it perfectly well, and if you will re-read my posts, that should be evident. There are two different primary network bottlenecks in SDV. One is the node. The other is the headend. The claim was made the most significant impact by TiVo's running suggestions would be at the headend, and that thousands of TiVos all recording channels 24 hours a day would swamp the headend server farm.

    To simplify things, we'll ignore services such as VOD and Video Rewind for the moment, and assume all the channels bear nothing but scheduled content. If the number of receivers on the node is less than or equal to the number of channels on the node, then channel blocking is impossible. Reducing the number of houses to that point is excessive, however, since utilization of channels on the node won't ordinarily reach 100%. First of all, as we all know, every program has a certain popularity, and a single channel may at times garner more than 50% of the viewing public. All those subscribers (or DVRs) viewing the same program receive the very same stream on a single carrier. This cuts down the number of unique receivers drastically and increases the numbe of homes which can be served by a single node. Most MSOs are targeting their network builds to wind up with between 500 and 1000 homes passed per node, which equates to something like 300 - 600 active subscribers and probably 750 - 1500 receivers. The number of active channels will tend to increase fairly rapidly at first with increasing numbers of receivers on the node, but soon the channel utilization begins to flatten out and only increases very slightly as more receivers are added.

    Here the number of multi-stream receivers such as DVRs does impact the number of active channels on the node, especially if things like suggestions are turned on, but the impact is only minimal. Again, it's not zero, but it's not huge, either, especailly once the number of available channels exceeds a moderate fraction of the total number of receivers. In this part of the network, biker is correct in saying the projections are usage based, but first of all, he was talking about congestion at the headend, not at the node, and secondly, the actual impact on traffic of Suggestions at the node is quite small. How small depends on how many channels are supported by the system, how many receivers are online, and of course the time of day. The thing is, at this point in the network, the Suggestions are going to have their biggest impact during the periods of lowest utilization, when blocking is least likely. That's not to say it can never happen. Since the node is considerably oversubscribed, blocking is indeed possible and in fact does happen from time to time. The odds of it hitting the same customer on a regular basis are exceptionally low, however. Unless the node is far too oversubscribed, no user should see a block network error on his receiver even once a month, and then not on any of the more popular channels. Indeed, the presence of Suggestions on TiVo's would tend to help prevent blocking on the scheduled channels while making it somewhat more common on VOD channels or during Video Rewind attempts.

    The headend is the second place where blocking can occur, but here the metric is different. At any node the network only has to be able to provide a sufficient number of streams to deliver all the channels being watched on that node, which could easily be fewer than 100. What happens across town is irrelevant. At the headend, however, there is no useage basis for delivery of service. It doesn't matter if one stream is being watched by 90% of the receivers in town. The load in the servers is dependent upon how many total streams are in service. If only a single viewer anywhere in the city is watching a particular program, it has the same impact on the headend as the one holding 90% of the subscribers. At the node the fact a single program is being viewed by a large fraction of the people on the node means the load on the node is potentially much lower. The impact of TiVos running suggestions at the headend is tiny compared with normal scheduled viewing and in fact is minuscule compared to the traffic generated by impulse traffic such as Video Rewind and VOD. Indeed, the traffic generated by all forms of scheduled viewing (including scheduled pay-per-view) is extremely small compared to the traffic generated by interactive video such as VOD and Video Rewind. The scheduled programs might represent 200 or 300 ir even eventually 500 or more unique data streams at the headend, but interactive services could easily eventually outstrip that by a factor of 100. There are a number of very clever schemes to allow the headends to handle such a vast volume, but the point is, if they can handle that volume, they can easily handle one or two or even ten additional streams required by having suggestions turned on.
     
  17. Dec 23, 2007 #857 of 2401
    lrhorer

    lrhorer New Member

    6,922
    0
    Aug 31, 2003
    San...
    Well, first of all the fact the channels here in San Antonio now exceed 1000 in numbering. Of course there are some significant gaps and 150 of those channels are audio stations with minimal picture content (more like screen savers), but still the actual number of channels is well over 500, most being pay-per-view and VOD. 72 are still analog. They have a 750MHz system here (they talked about increasing to 850MHz, but ultimately decided against it a couple of years ago). That leaves them about 230 MHz or 38 streams to deliver more than 400 channels, 20 or so of which are HD. I'm told all of their regularly scheduled digital channels offer Video Rewind, but I can't testify to that personally, because I have TiVos, and I left the company long before Video Rewind came out.

    I'm not allowed to give too many details, but we happen to share a number of facilities with them, and you wouldn't believe how much power they're using, or how much bandwidth for data. There's a reason for that.
     
  18. Dec 23, 2007 #858 of 2401
    lrhorer

    lrhorer New Member

    6,922
    0
    Aug 31, 2003
    San...
    Um, isn't that kind of what I said? A staged rollout would prevent their having to replace every cable box which didn't support MPEG-4 or to supply one to every customer without a CableCard device of some sort.
     
  19. Dec 23, 2007 #859 of 2401
    mikeyts

    mikeyts Stream Warrior

    2,389
    2
    Jul 10, 2004
    San Diego,...
    I'm sorry, but I seem to have missed something. Who is "we" and who are "them"?

    I wasn't aware that people are including the music services in the "500 channels" number. The bandwidth required for streaming music must be miniscule (and for feeding those screen savers, which I doubt are actual video streams). There are 64 of these channels on my system and I'd be surprised if they consumed more than half a QAM.
     
  20. Dec 23, 2007 #860 of 2401
    ah30k

    ah30k Active Member

    2,211
    0
    Jan 8, 2006
    My oh my, you do like to type a lot. I recall no claim that the bottleneck was in the headend. It may be here somewhere but I don't recall it. You laid out a scenario with 100,000 subscribers, tuners and 500 channels and claimed that SDV won't work because of the probability of more streams being needed was greater than 99%. That is where nodes fix the problem.

    I don't know why you are off on the headend bottleneck and 1000 interactive streams killing the system. That is a totally different topic. Perhaps you headend is just poorly designed.
     

Share This Page