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 23, 2007 #861 of 2401
    lew

    lew Active Member

    3,963
    3
    Mar 12, 2002
    I understand how SDV would allow a cable system to add more PPV channels and even VOD. I'm not sure how it lets them add more regular channels. The way I understand it if no customer on the node requests the channel then the channel won't be sent. What is the purpose of offering a channel that literally no one in a given area wants to watch? I guess it might make sense for a very niche channel, perhaps international.
     
  2. Dec 23, 2007 #862 of 2401
    dswallow

    dswallow Save the Moderatоr TCF Club

    50,977
    31
    Dec 3, 2000
    Long...
    Statistically there's always channels people aren't watching. By only actually sending video for channels that someone is watching, you can offer more channels than you actually have space to carry simultaneously. That's what Switched Digital Video (SDV) is all about. The efficiency exists because of the way subscribers are grouped and serviced in nodes. Every node has a maximum number of subscribers and receivers within it, and it's within each node that what channels are being watched that matters. The smaller the node, the more unwatched channels exist within the node.

    You can even think of the system much like a cell phone tower. When all available channels are being used simultaneously on a given cell tower, they break the cell up into smaller nodes, each serviced by their own tower.
     
  3. Dec 23, 2007 #863 of 2401
    ah30k

    ah30k Active Member

    2,211
    0
    Jan 8, 2006
    The key thing in how you phrase the question. It is NOT "Does anyone ever want to watch this channel?". The question is "Given the number of homes in this node, is someone watching this channel at this given instant?".

    Very different questions.

    I watch all kinds of channels, but only maybe two at any one time.
     
  4. Dec 23, 2007 #864 of 2401
    HiDefGator

    HiDefGator New Member

    1,851
    0
    Oct 12, 2004
    couldn't cable take the same route directv did and only do HD in mpeg4 first? then they wouldn't have to replace every cable box right away.
     
  5. Dec 23, 2007 #865 of 2401
    morac

    morac Cat God

    8,956
    22
    Mar 14, 2003
    NJ
    They'd still have to replace all the HD boxes out there. There's also the issue that there are no cable MPEG-4 boxes available currently. They won't be available till early 2008.

    I think eventually (years out) all new provided HD channels will be MPEG-4, but the cable companies can only use what's available now which is why HD is MPEG-2.
     
  6. Dec 24, 2007 #866 of 2401
    lrhorer

    lrhorer Active Member

    6,924
    0
    Aug 31, 2003
    San...
    "We" is the company for which I work. "Them" is Time Warner Cable.

    If you mean in terms of quantifying the performance of SDV, I would think "people" are excluding music channels, or at least they should be - which is why I specifically excluded them from my post. On the other hand, by either excluding the local FM stations in favor of the music stations or else putting the local FM stations on a QAM stream, they could gain back 20MHz or 3 QAM streams. That's a fair bit of content, and I suspect they may do just that as one of the first steps to going all digital.

    Compared to HD video, yeah. Compared to the $250 300 BAUD modem I bought back in 1983, it's downright screaming. ;)

    They're probably 200Kbps or so, so a QAM stream probably handles about 150 of them, so yeah, that's about right.
     
  7. Dec 24, 2007 #867 of 2401
    lrhorer

    lrhorer Active Member

    6,924
    0
    Aug 31, 2003
    San...
    'Not particularly, but I'm quite used to it. Writing technical documents including engineering briefs and detailed step-by-step instructions is a significant part of my job.

    See here:
    http://www.tivocommunity.com/tivo-vb/showthread.php?p=5787201#post5787201

    I never claimed any such thing. Quite to the contrary, my entire point during this discussion is SDV will work just fine, even with the tiny impact added by having large numbers of TiVos in the field with Suggestions turned on. The probability of more streams being needed is almost 0. The secenario I laid out is quite reasonable and only demonstrates that TiVo Suggestions won't cause issues at the CATV headend.

    I wasn't, and I never said it would, and it's not my headend in any case.
     
  8. Dec 24, 2007 #868 of 2401
    lrhorer

    lrhorer Active Member

    6,924
    0
    Aug 31, 2003
    San...
    Well, it's a big part of it, anyway. Another big part is the ability to deliver a myriad of interactive services.

    Your statement is correct, but I think it could be misconstrued. It's not quite what channels are being watched, per se, but rather that every node can have a potentially different set of channels being watched.

    Well, yes, but once again it doesn't matter how many channels are unwatched at the node, as long as the total number of channels requested doesn't exceed the total number available. The CATV provider's greatest dream is that every single channel on every single node would be active 24 hours a day. Anything less than that means some amount of theri investment is laying idle. Unfortunately for them, in order to maintain customer satisfaction, they are going to have to make certain the node utilization does not a number slightly lower than 100% for any significant amount of time during peak watching hours.
     
  9. Dec 24, 2007 #869 of 2401
    dswallow

    dswallow Save the Moderatоr TCF Club

    50,977
    31
    Dec 3, 2000
    Long...
    lrhorer, I think you're maybe just trying too hard in some of these replies. ;)

    The cable company isn't looking to have 100% utilization at all. All they want is to never have demand for 100.1% utilization or more. On a system-wide basis, they want to offer the greatest selection (i.e., be competitive to alternative services that may be available to customers). On a node level, they never want to have to tell a customer something isn't available. Every node doesn't have to ever see every available channel watched; certainly the cable company doesn't want to see weeks or months go by with not a single customer viewing any particular channel (as that would mean they could just drop the channel and save the money and nobody would even notice or care)... and with SDV they'll actually know if that ever happens, whereas with analog they'd never know.

    That balance is of course always subject to review, but it can be monitored very easily with the tools provided by SDV vendors they're using now. And as needed, they can split off nodes well before utilization reaches dangerous levels where they may face having to deny a customer access to a service they're trying to tune.

    As far as interactive services, they're not called Switched Digital services; they're called Interactive services. All that SDV has to do with them is simply efficiently using bandwidth for regular channels so that Interactive services have bandwidth for their own needs. The same goes for PPV or VOD services... no relation to SDV except that SDV enables them to have more bandwidth on the node available for such services.

    From many of the papers on SDV statistics it seems somewhere around 70% bandwidth utilization at peak times on a node is the "red flag" to initiate adjusting the node or splitting it completely.
     
  10. Dec 24, 2007 #870 of 2401
    ah30k

    ah30k Active Member

    2,211
    0
    Jan 8, 2006
    Doesn't the following quote mean you think that SDV won't work since only 1 of 500 channels isn't being watched?

     
  11. Dec 24, 2007 #871 of 2401
    mikeyts

    mikeyts Stream Warrior

    2,390
    2
    Jul 10, 2004
    San Diego,...
    lrhorer, I have to say that I too was getting the impression that you were starting to argue that SDV won't work, when it's been deployed in large systems for over a year and it is working (TWC Oceanic and TWC Austin alone are the 17th and 22nd largest cable systems in the country with 500K subs between them; your own system, TWC San Antonio is the 9th largest, with over 300K subs of its own). Are you using TiVo or a leased box? If you're using a leased box with access to the SDV groups, have you ever experience a denial when requesting an SDV channel?

    Also, you claimed that your system has over 1000 channels. I took a look at the TWC San Antonio channel line-up (can you tell that I'm bored :rolleyes:) and found:
    • 265 apparent linear SD channels, though there's some channels duplicated, assigned multiple numbers to put them in multiple ranges (i.e., the "Family Tier")
    • 77 "On Demand" channels (why is there a set of 9 channels--100, 192-199--marked "HD Movies On Demand"?)
    • 36 distinct linear HD channels (44, but 8 have two assigned channel numbers); 4 of these channels are the "other-coast" version of 4 premiums (HBO, Cinemax, Showtime and Starz)
    • 48 "Music Choice" channels
    • 63 total channels in PPV Events, PPV Movies, Movies On Demand (movies as pay-per-viewing-period VOD) and subscription sports-league tiers (i.e., "NBA League Pass", "NHL Center Ice", "ESPN Game Plan", etc).
    Altogether there seem to be 490 tunable channel numbers (on leased cable boxes subscribed to everything), with at least 20 channels with multiple assigned numbers because they appear in multiple tiers, so it's around 470 distinct channels. (That's the San Antonio line-up--I'm imagining that the Bandera, Blanco, Lake Hills and Stockdale line-ups are all smaller subsets).

    So, your system's channel line-up, while impressive (particularly in the number of VOD channels and especially the number of HD VOD channels), only approaches that 500 channel model, with plenty of PPV, subscription sports and VOD stuff in there and a ton of "other-coast" premiums (all innately switched VOD or highly switchable linears).

    EDIT: Sorry--you did not say that you system had more than a 1000 channels; in the post that I was thinking of (this, you said:
    Upon examination, that was bit off, but not nearly as wrong as what I thought you said :).
     
  12. Dec 24, 2007 #872 of 2401
    rictus

    rictus New Member

    17
    0
    Dec 28, 2006
    It seems like people are having difficulty understanding the state of the present argument, so here's what I understand so far.

    Someone (not lrhorer) was arguing that the existence of TiVos would cause SDV to fail because (1) the headend would get swamped with requests from TiVos tuned to random channels to record suggestions and (2) local nodes might also be getting lots of suggestions requests from TiVos.

    lrhorer is arguing that TiVos will *not* cause SDV to fail, because:

    (1) the impact of TiVos on the headend is small: pretty much every channel is being requested from the headend at any given time anyway (that was the point he was trying to make with his "500 channels" argument--he wasn't arguing that it would fail, he was arguing that this is true regardless of TiVos or SDV), and the number of HD TiVos being served by a single headend is likely to remain an insignificant proportion for some time to come in any case.

    (2) it's conceivable that there will be some issues at individual nodes, but they are likely to be infrequent, since again the proportion of TiVos among households served by a node is relatively small, and the times when TiVos are "wasting extra bandwidth" due to requesting suggestions are mostly at off-peak hours anyway (during on-peak hours, presumably a given household is likely to be watching/recording at least one show "for real", so would be considered in the calculations when deploying SDV).

    I don't know who's right--it seems to me to depend on how widespread TiVos are in a given area, although lrhorer is arguing that for any plausible scenario the numbers work out--but it seems reasonably clear to me what the argument is.
     
  13. Dec 24, 2007 #873 of 2401
    SCSIRAID

    SCSIRAID Active Member

    2,321
    0
    Feb 2, 2003
    Vero Beach, FL
    Your interpretation agrees with mine. lhorer has been arguing (successfully from my view) that SDV will not be negatively effected at the headend due to TiVo suggestions.
     
  14. Dec 24, 2007 #874 of 2401
    Luke M

    Luke M Member

    371
    0
    Nov 5, 2002
    One thing that's perfectly clear is that the worst case load of a Tivo - recording two switched channels in prime time - is exactly the same as for a cableco DVR. One Tivo user can't bring a switched video system to its knees. It's impossible.
     
  15. Dec 24, 2007 #875 of 2401
    mikeyts

    mikeyts Stream Warrior

    2,390
    2
    Jul 10, 2004
    San Diego,...
    Having looked over lrhorer's posts, I can see what he's been arguing, though I'm not sure what the assertion that a system with 100K subs, 270K receivers and 500 linear channels would get 100% of the channels requested during peak hours has to do with SDV, or the post by bicker he quoted in that response. It only takes one "5000 channel super-DVR recording every available channel 24 hours a day" to break SDV, since the premise is that the channels in SDV groups are assigned to over-provisioned pools of bandwidth (typically less than 50% as much as necessary to transmit all of them concurrently). A "super-DVR" sitting on one segment will break the system if it attempts to simultaneously request more of the channels in an SDV group than will fit in the bandwidth assigned to that group; any attempt to simultaneously tune all of them will definitely break it.

    Whether any bunch of TiVos on a given network segment recording suggestions would, by themselves, simultaneously request too many SDV channels is another topic. It seems highly unlikely, but creating recordings of suggestions, all of which have a very low probability of actually being viewed, is just not an important enough activity to allocate any portion of a precious shared network resource to. No properly planned SDV group should contain any really popular services, so any 500 channel system should contain a couple of hundred non-SDV services to record suggestions from, if you simply must :).
     
  16. Dec 26, 2007 #876 of 2401
    lrhorer

    lrhorer Active Member

    6,924
    0
    Aug 31, 2003
    San...
    'Not even. I'm flabbergasted anyone would get that impression. I was rebutting the notion that TiVo Suggestions breaks the SDV model and would cause a severe problem for CATV providers and so TiVo should limit the scope of its Suggestions feature. The notion is nonsense. Ignoring the legal and logistical mess we're in with a lack of unified standards and the inability of any TV / DVR manufacturer to hedge their bets against losing money by picking one of the flavors of SDV, SDV already provides a vastly expanded set of features with far, far more to come. Exactly what statement in any of my posts would have led you to believe I was arguing SDV would not work? It works very well , indeed. It works substantially just as well with myriads of TiVos with Suggestions turned on as without.

    TiVos, but that's beside the point. I worked on early trials of SDV and other technologies when I was an Engineer for the company which was bought by TWC, and I still have several friends in the Engineering department there. Denial requests are uncommon, and TiVos won't change that appreciably, suggestions or not. Denials are far more likely to be produced by and experienced by non-scheduled event requests which leaves the TiVo mostly out of the equation.

    No, I said the channel numbers exceed 1000. In fact, the highest channel number is 1310. I also pointed out there were substantial gaps.

    Uh, yeah. No way was I going to do a count of the channels.

    Each channel has a different set of movies available. The VOD interface is clumsy, so to help they divide it up into different channel numbers. The VOD interface needs lots of work, if you ask me.

    The point being? Note classical Pay-Per-View venues are still scheduled. They require authorization, but they starta sn stop at a specific time. IF one person in the city requests the PPV event, then every node in the city can request it without additional impact at the headend, and if one subscriber on the node has requested it then every subscriber on the node can request it without additional impact at the node level.

    In any case, the fact is any VOD "channel" is not a single channel, but rather an interactive stream, and two different individuals requesting the same program 5 minutes apart are NOT using the same data stream, but an entirely different one - effectively a separate channel. Thus, as I already pointed out, the load on the CATV headend from those two lone subscribers is as high as that of several thousand TiVos with suggestions turned on. Out of 300,000 subscribers, the number of data streams from people ordering VOD offerings and / or requesting Video Rewind will far exceed the number of scheduled programs being requested, which includes all TiVos running Suggestions. At the node, it's unlikely the two subscribers in question will both be on the same node, so the impact from VOD and Video Rewind on an individual node is much less than at the headend. Conversely, the impact from a TiVo running suggestions is much higher at the node level than at the headend (where it is virtually zero), but still quite manageable and quite small. Even if every single household had Series III TiVos with suggestions turned on, the CATV provider would not have to increase the number of nodes by even 5% or 6% - if that - and it's going to be a very long time indeed before TiVo penetration in the CATV subscriber population even reaches 10%, let alone 100%.

    That's right, I didn't. I also didn't say SDV would not work.

    To sum up what I DID say, the maximum possible load both at the headend and at any node produced by TiVos running suggestions is equal to the total number of scheduled channels. which in my local system is less than 200, and in my hypothetical system is 500. The number of "channels" the headend must produce for VOD, Video Rewind, and other interactive services is in the 10s of thousands. If the headend can handle that, handing all the TiVo suggestions on Earth is trivial. At the node, Suggestions do increase the load very slightly, but only to speak of amnd mostly in off-peak hours when increasing the load is not a problem. Is it possible one or two of the 300,000 subscribers a day might experience a denial they otherwise would not have experienced if there weren't a bunch of TiVos on his node? Yeah, maybe, but that averages out to maybe once a year per subscriber, and that only once the number of TiVos running Suggestions exceeds 20% - 50% of the customer base. At the rate Series III TiVos are selling, that's going to be a couple of decades, at least.
     
  17. Dec 26, 2007 #877 of 2401
    lrhorer

    lrhorer Active Member

    6,924
    0
    Aug 31, 2003
    San...
    No, it doesn't (at least at the headend) and that's the point I'm trying to make. It doesn't matter if there are millions of 2 channel DVRs or thousands of 5000 channel DVRs, All put togethr they can only request 500 channels, because that's all there are. Now yes, that single 5000 channel DVR could swamp the node its on, if it actually tried to request all 500 channels, because the node can't handle the entire linear bandwidth, let alone adding in the interactive requests on the node.

    How low is "low"? Again, when we are talking about DVRs in general and TiVo Suggestions specifically, we are talking about a scheduled event. It could be PPV, but it isn't VOD and it isn't Video Rewind or any other interactive service. A total of 500 scheduled channels is very large, indeed, but lets say they wind up with 1000 scheduled channels and 5000 interactive events. In order not to be wiewed by at least 1 subscriber, the event is going to have to have a less than 1 in 300 share. Given 1000 channels broadcasting during primetime, every idle stream must have a less that 1 in 300 market share during primetime. Does anyone really believe any significant number of broadcasters are going to schedule a show during primetime which has a less than .34% rating? The notion is just silly, but even if a bunch of broadcasters did schedule a bunch of lost causes during primetime, it still would not even come within several orders of magnitude of the number of streams whihc could be requested by interactive services like VOD and ezpecially Video Rewind.

    Well, eventually it will be essentially 100% SDV, except for basic service. After all, it doesn't cost the CATV provider anything more to deliver a channel as SDV than not. It's the same equipment producing the same streams. The differentiation is not between SDV and linear, but between scheduled and interactive. The DVR - Suggestions or not - has a limited impact on the number of scheduled programs being viewed at the node level and zero at the headend. It has zero impact period on interactive services.
     
  18. Dec 26, 2007 #878 of 2401
    lrhorer

    lrhorer Active Member

    6,924
    0
    Aug 31, 2003
    San...
    Right. The other piece of the puzzle is that the number of subscribers on the node is quite small, comparatively speaking. A program only watched by 1 person in 1000 is still going to be watched by 300 households in a 300,000 subscriber system, but those 300 households are only going to impact roughly 300 nodes, and the other 300 nodes in the system can use that same bandwidth to transmit something else - or a bunch of something elses.
     
  19. Dec 26, 2007 #879 of 2401
    Luke M

    Luke M Member

    371
    0
    Nov 5, 2002
    It certainly does cost more. The switching function isn't free, nor are QAM modulators free.
     
  20. Dec 26, 2007 #880 of 2401
    dswallow

    dswallow Save the Moderatоr TCF Club

    50,977
    31
    Dec 3, 2000
    Long...
    The signal has to be encoded and modulated no matter what. So while there might be a slightly higher cost for a frequency agile configuration or to be able to divert the signal to any of many different encoder/modulators, the cost is still there.

    Once you've invested in the technology, there's little reason not to fully utilize it. Nodes will be small enough that even popular channels won't be viewed by anyone at some time. And the bandwidth that would otherwise be wasted to carry a channel 100% of the time when it is needed perhaps only 90% of the time does have value.

    Though it's still the use of any channels to carry analog signals that's starving cable systems of bandwidth needed to handle more SD and HD channels. SDV wouldn't be necessary anywhere yet if analog could go away completely... or even be reduced to just 15 or so channels.
     

Share This Page