Calculating the Energy Cost of Moving DVRs into the Cloud

How much energy would it cost to move all of our digital video recording into the cloud? If the cable companies all followed Cablevision’s path, it would require an estimated 300 megawatts of power, or about one third the output of a nuclear power plant.

At an industry event presented by the Society for Cable and Technology Engineers (SCTE), Comcast’s Mark Coblitz today highlighted one unintended consequence of the Remote Storage DVR (RS-DVR) system Cablevision uses for its cloud-based video recording service. Because of legal rights issues, the MSO has to create and store a unique copy of every program recorded in the cloud by its subscribers. According to Coblitz, if the industry did the same thing as a substitute for roughly 30 million home-based DVRs, energy waste would go through the roof. In contrast to the 300 megawatt number Coblitz cited, a network DVR system keeping just one copy of up to two million different programs would require only about five megawatts of power in total.

The cable industry is examining power consumption issues a lot more closely these days. It’s not so much a matter of environmental responsibility, as it is the financial bottom line. If more operators start to migrate toward network-based recording, you can bet they’ll also begin to exert serious pressure in court rooms and board rooms to move away from the RS-DVR model. True nDVR would be a lot more efficient and environmentally sustainable. More importantly for MSOs, it would be a lot less expensive.

11 thoughts on “Calculating the Energy Cost of Moving DVRs into the Cloud”

  1. It’s so stupid that the legal crap keeps this from happening.

    It would make so much more sense if instead of all of us having a HD under each of our TVs we just communicated with a central server via SDV and got a show we wanted to watch, commercials and all. It would even eliminate the scheduling nonsense and just put us into the notion of Favorite Shows.

    The media companies just don’t get it in regards to this or in regards to DRM-free content (I want to be able to download DRM free movies from Amazon or iTunes – instead I buy DVDs – different subject, though).

  2. “It would make so much more sense if instead of all of us having a HD under each of our TVs we just communicated with a central server via SDV and got a show we wanted to watch, commercials and all.”

    My bolding.

    As a consumer who has a happy relationship with an owned and operated DVR, count me as one who’s very happy the vagaries of the laws have prevented “cloud DVR’s”.

    I like being able to comskip. I like having control over content windowing issues under my management. “Cloudiness” means loss of end-user control. Local caching for the win! (No one will miss local DVR’s until they’re gone. Then they’ll be much gnashing of teeth and pulling of hair.)

  3. Sorry, I was thinking in terms of:

    The recording taken from the local broadcaster or the normal satellite feed (well, with local commercials inserted – just as if you recorded at home) but is just stored somewhere else.

    I would want to retain the ability to skip commercials as well although you could definitely see them thwarting that (which is kind of dumb as they take a good idea and make it worse).

    Basically I’d love to have a well-done DVR (like TiVo) but instead of it looking on a hard drive in the box it’s actually looking out at a server somewhere and everything else remains the same, functionally. That server should just store one copy of each channel, not a separate copy per customer.

  4. Even if the operators have to continue to store a unique copy for each user they can still use commercially available technology to solve the power problem.

    I could see four or more ways of doing this:

    1. Even if they’re required to keep a copy for every person, it is common industry practice to deduplicate data at the storage layer for long term archival data like this. Though, deduplication could be argued against as not following the letter of the law. The less data is stored, the less power is used.

    2. Even without deduplication you could just store all but one copy on non-spinning media (i.e. tape consumes no power at rest). Then only serve up the one copy left on disk.

    3. If the store many, serve one practice is rejected then they could use something like a massive array of idle disks (MAID technology) where only a small percentage of the disks are ever powered up at one time.

    4. A more complex hierarchical setup can be used where only a stub of the data is kept on active disks and the remainder is on tape (again, no power for tapes not in use). When the user presses play the stub is large enough to cover the load and transfer time of the tapes.

    These existing technologies could easily be deployed to solve the operators’ power problems and make network DVRs a reality. Though, you’ll have to pry my TiVo from my cold dead hands.

  5. I’d bet the RS-DVR storing individual copies for every user is still more energy efficient than individual DVRs in each home. Just from economies of scale. Industrial drive arrays are more efficient per byte, and there are also efficiencies is power conversion, cooling, etc.

    But still wasteful compared to an optimized solution with just one copy shared among multiple users.

  6. I agree with Chucky… the cloud is great in theory… In reality it takes control away from the end user, and in general reduces security for the smart end user. I’m smart enough to advance schedule and buy extra storage.

  7. I think the dedupe process would meet the letter of the law and the real record keeping is “did the requestor set this show to record?” if yes then serve up the deduped copy.

    The ‘legal crap’ however is the content owner wanting to keep to some schedule. They likely are afraid they can not sell the advertising that pays them to be content makers if people do not watch in real time. This is not a real fear in the long run but the solution would indeed mean we lose control of comskip and the MSO simply inserts timely commercials into the stream. This cloud DVR approach could mean we all go full circle back to watching commercials, but we can start the show when we wnat and can pause it. For the right price I can be on board with that but I much prefer my current ability to record shows as I want and watch them as I want, even if the cable is out.

  8. Thanks anyway, but I’ll keep my local, Tivo, DVR to record shows off-the-air, supplemented by shows bought via Amazon.

    No DRM worries for me…

  9. I worked on the product and know something about the legal angles. The case succeeded because the RS-DVR was logically identical with a DVR in the home. No due duplication would have passed muster.

    That said the question is how far can you go without getting sued? I assume over time the line will be successfully pushed farther with new deployments. But it will take a LONG time.

    I presume the obvious path is to deduplicate shows from networks you get agreements with. Perhaps the carrot would be updating the ads if the show is played after three days…

    Yes they’ll obviously block you skipping ads or they will once people don’t have regular DVRs anymore.

    Like Megazone I think they left out the part where the current DVRs are even less efficient than an RS-DVR, deduplication or not…

  10. I can understand resistance to the “Did the user set this to record” bit scheme. The logical power utilization argument can be used in this case, but once the solution is up and running, the next feature requested will be a “Just record everything” option, as storage of the video would no longer be an issue.

Comments are closed.