It’s mostly a privacy issue. For example, if Netflix used P2P, you’d be able to track which IPs had parts of which movies and start profiling Netflix customers. That would be bad.
The problem with P2P is that you have to make what you have public. That’s fine for a game update where it’s all the same, but for movies there are huge privacy implications.
The issue with using it for Netflix isn't privacy, it's that bittorrent is simply not any good for streaming video content when compared with current technologies already used by streaming broadcasters.
* It's slower to deliver video since you're reliant on upload connections of end users
* It's less reliable (same reason as above)
* You have no guarantee that chunks will arrive in the right order to actually stream data (ie the end of the file might download first)
* You cannot send multiple different video qualities in the same stream and have the client dynamically pick the right bitrate for their connection (this is what current streaming services do and why you often see Netflix / Prime video / etc switch between quality mid-stream without having to restart that stream. I can write more in this if people are interested)
* It's harder to debug network problems if you are experiencing issues with video quality (been in enough stressful emergency meetings with network guys over the years - I can't imagine how much worse it might have been if we couldn't do a full end to end trace of the delivery)
* Time to start playing a new stream is longer (which end users might care about)
* It couldn't support live services where the video data is being generated near real time
The current methods for video delivery are actually really good, bittorrent would be a major step backwards. However for delivering other kinds of assets - such as patches to computer games - protocols conceptually similar to bittorrent are used.
> It's slower to deliver video since you're reliant on upload connections of end users
Doesn't really matter, it is fast enough. If it isn't you back it up from a datacenter.
> It's less reliable (same reason as above)
Shouldn't be any real difference in reliability.
> You cannot send multiple different video qualities in the same stream and have the client dynamically pick the right bitrate for their connection (this is what current streaming services do and why you often see Netflix / Prime video / etc switch between quality mid-stream without having to restart that stream. I can write more in this if people are interested)
Sure you can. You only have to have a small map between play time and file offset for the different streams, the client will then pick whatever it wants.
> You have no guarantee that chunks will arrive in the right order to actually stream data (ie the end of the file might download first)
The client can decide this. There are torrent clients that do this already. Buffer 2 minutes, if some chunk is missing when 20 seconds remains pull it from a datacenter instead.
> Time to start playing a new stream is longer (which end users might care about)
No, you start streaming from the datacenter.
> It couldn't support live services where the video data is being generated near real time
Not sure, but if a minute is acceptable delay (depends on what is being broadcasted) it should be feasible. A live webcam for a tourist resort should be fine, a sports event, maybe not.
Spotify even used to work this way. When a user clicked play (or seeked to a different part of the song) the first (if I remember correctly) 15 seconds was fetched from their CDN. After that it used its own torrent-like system to continue and pre-fetch the next song.
> Doesn't really matter, it is fast enough. If it isn't you back it up from a datacenter.
Sorry but no it really isn't. You'd have to rely so much in your own data centre that you'd loose the benefit of bittorrent. Plus many services will use a commercial CDN, none of which current support bittorrent. So you'd end up having to build your own infrastructure there, which would be a great deal more expensive.
> Shouldn't be any real difference in reliability.
Home connections might drop off due to power cuts, router failure or any of the other numerous conditions that datacenters battle against. Home connections might get throttled by ISPs.
Where I work we offer 5 "9s" of reliability on some services, good luck asking consumer broadband to offer the same ;)
> The client can decide this. There are torrent clients that do this already. Buffer 2 minutes, if some chunk is missing when 20 seconds remains pull it from a datacenter instead.
2 minutes is an excessively long buffer compared to.how long most RMTP segments are (often 15 seconds or below).
> No, you start streaming from the datacenter.
So then why bother with bittorrent at all?
> Not sure, but if a minute is acceptable delay (depends on what is being broadcasted) it should be feasible. A live webcam for a tourist resort should be fine, a sports event, maybe not.
A minute isn't even close to acceptable delay. Imagine if you're watching a sports match and Twitter is ahead of your video fees, you wouldn't be grateful for the spoilers. This is a real world problem with video streaming and why they talk about getting the latency down to 5 seconds or less.
> Spotify even used to work this way. When a user clicked play (or seeked to a different part of the song) the first (if I remember correctly) 15 seconds was fetched from their CDN. After that it used its own torrent-like system to continue and pre-fetch the next song.
Spotify is audio only. People love to compare video streaming to audio streaming but they don't realise that HD video is an order of magnitude more complex to stream - in terms of bandwidth, syncing, dropped frames, etc.
It's one of those problems that seems easy to solve from a superficial level but when you start getting to the broadcaster level it's actually a great deal more complex than even Plex and other self-hosted streaming services would lead you to believe. (Disclaimer, I've worked at the broadcaster level)
I just can't address this further. I'm not convinced you have the slightest clue what bittorrent is (how on earth does not having 5 9s matter for a consumer connection? If a node drops noone will notice).
I'm not saying there won't be problems. A major problem is the asymmetric nature of many consumer connections. Not only is the upload often a fraction of the download (that is the easy part), but the download speed can be greatly sacrificed if upload is utilized. Add to that issues that home users might want to use the connection for other stuff.
The nightmare and confusion surrounding "I can't game because someone is watching torrent-tube" will be real and add to that issues with ISPs that have a fixed quota or people on mobile (or tethering a mobile connection to laptop). Netflix et al would not like to deal with that kind of FUD spreading around.
And all this is before considering local ISP bottlenecks as it isn't what the network was designed for. A vastly superior option is to put a proxy on the ISP network itself (but yes, hard to do for small players).
Those alone are probably a magnitude worse than the technical issues you speak of. Even spotify got much flak for it, and no, I don't think anyone on earth think the bandwidth requirements of audio and HD video are similar.
I actually did cover that point (albeit at a much higher level) and of course I know what bittorrent is. Just because I don't agree with you it doesn't mean I don't understand the problems. Quite the opposite, I happen to work in the video delivery industry so it's my job to understand these problems.
Like I said, it's ever so easy being an armchair critic but try deploying this stuff at scale with SLAs covering 5 "9s" and paying customers then tell me you've got all the kinks in bittorrent solved ;)
Arguably, no, you contrive issues that don't make any sense. Such as presenting the issue that the end of the file might be downloaded first. Or that you can't have different quality versions and dynamically switch.
The end will only be downloaded if the client requests it, which won't be unless the user is watching the credits...
The client matters though. Even just getting that client onto your end users systems is going to be problematic. And even those bullet point aside I raised other issues you're yet to reconcile. Plus there are other problems I'd not considered (eg RMTP is often sent over UDP and bittorrent is not only TCP but has additional CRC checks that you don't really want with video streaming).
But as you're clearly an expert in video delivery you should build a torrent based video delivery platform. You could make a killing (assun. Or alternatively you might discover these opinions you were dismissing might have had a point and actually video delivery is a lot harder than you first assumed.
I know it can be done via bittorrent. Anything can be hacked to do anything if you're motivated enough. However the question asked was whether it's good enough for paying customers in a commercial environment (ie Netflix). That's a very different set of expectations.
Popcorn time when I looked at it has consistently given me performance at least as good as Netflix (which tends to downgrade quality for bad reasons - I have very fast fiber and my ISP has good peering - and has a client that can't seem to stop desyncing audio). The only reason not to use it is arguable legality and the fact that content creators deserve money. It's definitely not quality of service.
If you're on a really fast internet connection where your ISP doesn't throttle P2P (many do in the UK) then I wouldn't be surprised that you'd get good results from Popcorn Time. The issue is whether that is consistently true for people on slower internet connections, connections with QoS or other traffic shaping methods against P2P, and those who stream on lower end hardware (old laptops / phones) or embedded devices (Amazon Firestick or smart TVs).
This is where RMTP streams really shine through - they offer the performance (at the client level) and flexibility (quick swapping between bitrates within 15 seconds of chunked video) to maximize the video quality per client.
As for why Netflix doesn't look to great on your connection, I can only speculate there but the reasons might be more down to congestion on your local Netflix cache server or your ISP itself than RMTP streams being inferior to BitTorrent.
I have a good provider that gives me what I pay for, yes. I understand that many countries have a borderline third-world internet situation, but that has very little to do with the problem you were presenting. What you're talking about here is essentially that Netflix has the lobbying power to make sure that it's services are put above other services in countries vulnerable to that kind of manipulation. I believe that instantly.
I can understand what you're talking about, but I think it's important to appreciate that as an end user I don't really care about what tech words I can tell myself about why things look bad. I care that one thing looks bad and the other does not.
I don't think you understand point. I'm not just talking about "borderline third-world internet" situations. I'm talking about watching stuff on embedded devices, on public WiFi, on older harder or on your cell phone while out and about. I'm also not talking about Netflix's lobbying power.
My point is quite simply that RMTP is already a better protocol for video delivery than BT.
Yes Popcorn Time exist and in some situations it is comparable to Netflix, but not in all situations. In fact not even in most situations.
BT wouldn't allow you to deliver to other providers such as YouTube or Twitch (this is something I was working on last week with RMTP streams). BT wouldn't allow you to dynamically inject ads separate to the video feed (this means you'd need to encode your adverts into the video file so you cannot charge different rates for different days or viewing times - which is a deal breaker for most broadcasters). BT wouldn't work for live feeds (so it would be useless for sporting events - which is what generates a large chunk of broadcasting revenue). BT is noisier than RMTP at the ISP level (ie it actually costs ISP more bandwidth not less - and given how much many ISPs are already complaining about Netflix et al, the last thing you need to is expect those ISPs to offer Netflix free bandwidth in for the form of user seeding).
As you can see the point I was making right from the start isn't that you cannot stream video via BT but rather that there are alreadybetter protocols for video delivery than BitTorrent. Hence why professional video delivery platforms don't use BT.
I suspect the issue many commenters on here have is they assume that traditional video delivery is still a classic straight stream of data like the old days and like people are familiar with when downloading via HTTP or FTP. But that's simply not how RMTP works. Modern video feeds are actually chunked just like BT is, except that it's feed from a CDN rather than peered from end users.
Client matters, yes, that hasn't even been brought up in the discussion.
Never stated I was an expert in torrent or broadcasting. But barely one who has used bittorrent can counter most of your arguments. As I've mentioned there are tons of issues (and as mentioned elsewhere in the thread, privacy alone would make it a non-starter for many). I don't believe in it for large scale, but not for the technical issues you presented.
Edit:
Apologies if I came off too harsh, been a rough couple of days.
They haven't explicitly 'disproved' all of your points because as mentioned previously most of what you said is nonsense.
(Someone already mentioned this, but seriously - this has been done - it works really well - as a polished commercial product it would probably work even better: https://en.wikipedia.org/wiki/Popcorn_Time)
I've already addressed Popcorn time several times in this discussion.
The fact that you've acknowledged that it's not as polished as a commercial product should be a big hint that perhaps commercial products choose RMTP over BT for a valid reason. Perhaps even for the reasons you initially dismissed as nonsense. Perhaps you might need to read up a little more about how professional video deployment actually works before you assuming you understand video deployment better than all those engineers who do this shit for a living. Just because something conceptually works it doesn't mean it's any good when dealing with the expectations of paying customers who might want SLAs of 99.999% uptime, low latency live services (like less than 10 seconds), near zero buffering times etc.
The Dunning–Kruger effect is overwhelming in this thread but trust me when I say video engineers are not stupid people and if their lives could be made easier by using BT then we would definitely be seeing commercial uses of BT for video deployment.
I thought I pretty much disproved every point you made. Here are some further comments but honestly the issues you bring up doesn't make much sense.
> If you have to rely so much in your own data centre
Not sure if we are talking 15 second gifs or full-length films here. The thread started with netflix as an example and that's the context I've assumed. The startup-cost on a full-feature film or even a series episode is barely a fraction in the context so I really don't get your comment here.
But I would go even further and expect constant backing from your data center to help with distribution, depending on how incentivized people are to keep the client alive after they've seen something. I would also not, both for network performance and common decency+courtesy, try to squeeze everything out of my customers internet connections.
> Home connections might drop off due to power cuts, router failure or any of the other numerous conditions that datacenters battle against. Home connections might get throttled by ISPs.
None of that applies here, that's a huge point of bittorrent to be resilient against that so it just doesn't make any sense to bring it up. A datacenter is centralized so it needs to deal with it. Thousands of clients spread throughout the world will likely even have better reliability than a datacenter. All in all, no reliability lost. Throttled by ISPs are a real issue and I really highlighted those issues earlier.
> 2 minutes is an excessively long buffer compared to.how long most RMTP segments are (often 15 seconds or below).
I just made that number up, but yes, a torrent solution will need a larger buffer. That really isn't a problem in this context. What is a much bigger problem is the storage needed for seeding depending on how you want to solve that and how reliant you want to be on your clients (again, your points doesn't make sense, why complain about the buffer when the cache is orders of magnitude worse).
> So then why bother with bittorrent at all?
I wouldn't, and I never said I would. But the protocol would work for it, consumer ISPs and other factors might not for such a large scale operation. And the privacy implications are also bad, it would leak information to your competitors as well. Bandwidth costs clearly aren't that big of a deal either so the potential gain isn't worth it (talking netflix scale here). Also proxys at ISPs are a superior solution.
> A minute isn't even close to acceptable delay[...]
You can not possibly have read what I wrote.
> A bigger issue is just getting content owners on board to begin with
Content owners are notoriously irrational so I wouldn't argue against that. But torrents could still be encrypted.
I suspect some of the issues you're having understanding my comments is I've assumed you had a working knowledge of RMTP.
Basically the protocol comprises of multiple streams of various bit rates. These streams are chunked into blocks of n seconds. This is often around 15 seconds or less. The client will hot swap between each stream if one chunk takes too long or downloads too quick. This ensures that the video quality dynamically scales to the best bitrate your bandwidth will handle and it does so quickly and with minimal buffering. It works for video on demand services but it also works for live feeds as well and the whole thing can be served over multicast or UDP for further savings at either the DC or across the wire.
Bittorrent is nowhere near optimised enough to compete on that level. The 2minute buffering you guessed at would easily look terrible in side by side comparisons with RMTP deliveries. Not to mention a whole boat load of hardware encoders (rack mountable gear used in broadcasters) are built for spitting out RTMP streams yet none (at least that I'm aware of) exist for BT. So you'd have to build your own gear as well as content delivery network. That alone would cost you far more than any savings P2P deployment might earn you.
That is a good point. I still argue you can do the same but with worse results with a torrent solution.
If you have 2 minutes of buffer but realize you can't keep it up you can still switch to a lower bitrate, that will allow you to get any missing chunks and maintain the buffer going forward. So you can still adapt (from the networks point of view much quicker than the 2 minutes), but not nearly as fast and the end result (quality change) might lag quite a bit.
> Never stated I was an expert in torrent or broadcasting. But barely one who has used bittorrent can counter most of your arguments.
The way you discussed this came off as incredibly arrogant, so you shouldn't hold your hands up all of a sudden and claim you're being misrepresented—I personally expected you were a bittorrent developer disgruntled over lack of adoption for non-technical reasons or similar based on the vitriol and self-assuredness.
There is a way to point out that some of those things should not theoretically be a problem without acting like they have no practical merit whatsoever and that anyone who believes so is an idiot. When you cite audio streaming you especially make yourself look like a Dunning-Kruger case as delivering adaptive HD video involves many complex issues that are completely non-existent or trivial for audio. It's hard enough with full control over your own PoPs, and layering on P2P only makes it more complex. Content rights / privacy aside, you could never rely on this as primary delivery for commercial content, so in effect you'd be layering a bittorrent system on top of a traditional CDN setup as a cost saving measure, but you've have to be damn sure QoS didn't degrade in the process which is also non-trivial. Just because it's theoretically possible doesn't mean it's a good idea.
There is a way to point out that some of those things should not theoretically be a problem without acting like they have no practical merit whatsoever and that anyone who believes so is an idiot.
I felt the original message had the same tone and thought to answer with the same. I shouldn't have. And I went further. And I just couldn't stop. I shouldn't be commenting when I'm this frustrated (unrelated to this thread).
Nah you're fine. I'm getting serious "ideological/political argument" vibes off the other commenter.
I recognize the other commenters tactics as the same "throw a bunch of dubious claims at the wall, force your opponent to spend way more time refuting each one than it took you to make them, and as soon as they look like they're climbing out move the goalposts again" thing that I see in political arguments with half drunk friends and strangers on political subreddits :P
I'm sure I've seen a great webcomic that describes this somewhere but I can't find it...
Your only point was popcorn time and that stacks up poorly against a good RMTP stream.
The problem we have here is that a bunch of hackers have seen it can be done with BT (which nobody is saying it can't be) and then assuming that BT is suitable for someone like Netflix. If it were that clear cut then we would already be using it. But the fact remains BT doesn't actually stack up that good against already established technologies we use for video streaming.
I'm not saying this for the sake of an "ideological/political argument", this is a well researched point I've discovered doing my professional day job.
It's ever so easy to hack something together that works for a bunch for non-payment customers who just want to pirate something. It's a whole other ball game to build a production service with advertisers (where applicable) and other partners and sell that to paying customers.
What you're doing is comparing a kit car to a bus and saying they're the same because they both transport people. Then assuming the kit car can transport the 40 people at a time for 12 hours a day every day because you've made the previous logical assumption of equivalence.
To your first point on speed: check out Popcorn Time. Even though this example is, likely, streaming copyrighted material - it showcases that BitTorrent is indeed fast enough to stream reliably at FHD quality. Bandwidth, even from a mildly seeded torrent, can easily push down hundreds of megabits per second. As others have noted place a few well connected, data center oriented, seeds and you have speed and distributed sources (reliability). Content caching boxes already handle the majority of Netflix traffic at the DC/head end of your ISP today. So Netflix could easily add BT to that caching layer and bake it in to their native apps (Android/IOS) as a first test bed.
If you're placing cache servers at the ISP then they will charge Netflix for the seeding bandwidth, which would work out more expensive than running traditional RMTP streams because the BT protocol has much more overhead (TCP vs UDP, CRC checks, tracker syncing, etc). And that's without factoring in the likelihood that the cost of bandwidth per GB from a CDN would be much lower than from an ISP.
So what would Netflix gain in doing so? They'd need to build new clients, convince content owners that P2P is secure[1], convince ISPs to seed content, pay for new infrastructure, etc. And the final product would be more expensive and _at best_ equivalent to what they already have - though realistically it will be worse on all but the top end viewing platforms (ie laptops and desktops). And for added bonus this infrastructure "upgrade" then removes the possibility of Netflix running live services and dynamic ad injection, etc.
So it's more expensive, less future proof, less performant on low end internet connections and wouldn't work on any of the existing smart TVs or streaming HDMI sticks.
Like I've said in other comments, BT is fine for hackers and pirates - for people who want to run something for free, willing to run it on a laptop or even willing to be inconvenienced a little for the sake of running something different. But it simply couldn't compete once you start scaling it out as a commercial platform with all of the customer expectations from non-techies and the unseen requirements / complications that happen with professional video delivery platforms behind the scenes. Plus for all of this talk about how good BT could be, people seem to miss the key point I made right from the start: that existing delivery platforms actually aren't that bad.
[1] I actually have no qualms with the security side of P2P in terms of protecting copyrighted content. But content owners are notoriously slow to adapt to change and hard to please. Getting them on side might prove an even tougher problem than engineering BT to compete with RMTP at the commercial scale.
RMTP license is pretty interesting as it among other things prohibits circumvention of "technological measures for the protection of audio, video and/or data content"
Also things like ad injection are not features for the end user.
RMTP is a proprietary protocol developed by Adobe so it's a good fit when you want to screw your customer with DRM and other stuff.
I don't disagree with your points however it's content owners who decided if content should be DRMed or not, not the end user. Thus you'd find videos delivered via any other protocol (Inc Bittorrent, if that ever became a commercial platform) would also be subject to DRM.
Similar point for as injection. You might consider it anti-user however in many cases ads are what pays the bills and thus in traditional "over the air" broadcastering it's actually more important to avoid down time during as as spots than it is during a programme airing because the cost of compensation is greater.
Ultimately in professional video delivery you'd have more than one type of customer; obviously you have the consumers watching the content but there is also the people paying for ad slots. Plus in many cases you wouldn't be the content owner so you'd have a third customer who is the person paying for your services to deliver the video.
So the features you might consider to be anti-user (and to an extent I don't disagree with you) are still there for paying customers. Or to out things a different way, video engineers would prefer not to add the complication of DRM and as injection into their infrastructure if they didn't have to.
As for RMTP being Adobe owned, there are other video delivery protocols around but RMTP tends to be the preferred platform for B to C (broadcaster to consumer). Or at least that is the case in the places I've worked. Eg it wouldnt surprise me if Apple / iTunes did their own thing.
Your argument is completely valid in the current context, which is the only that matters. Just FYI, In other contexts these points are more nuanced:
> It's slower to deliver video since you're reliant on upload connections of end users.
This has been scaling really well for small unfunded projects that need to deliver big binaries, such as open source distributions.
> It's less reliable (same reason as above)
If you have large corporate or state level actors trying to censor the content, then this becomes the more reliable method.
> You have no guarantee that chunks will arrive in the right order to actually stream data (ie the end of the file might download first).
This is not fixed by the protocol, but by clients optimizing for total download speed. I think some clients allow you to download more sequentially.
> You cannot send multiple different video qualities...
Video torrents often come in 720p, 1K and sometimes even 4K versions. So you could pick in advance. But that's not dynamic mid-stream like you mention.
Since you offered to write more: Is there such a thing as a progressive codec for audio/video? I vaguely remember video chat codecs doing something like this, where you get progressive degradation if you loose packets.
> It's harder to debug network problems...
oh yes...
I think this is the main reason Skype and Spotify moved away from the P2P tech that they used initially. I hope this becomes less of an issue when projects like LibP2P and IPFS mature and become as reliable and common as TCP/IP and TLS.
> Time to start playing a new stream is longer...
You can solve many of the issues mentioned above by having high quality sponsored seed servers. This is basically a hybrid approach. The seeds would guarantee good services and the P2P part would reduce load on the seed servers where possible.
> It couldn't support live services where the video data is being generated near real time
Not Torrent, but there's community of P2P television streams, quite popular for big sports matches. And from what I hear great quality. This is a completely different P2P model, more like a multicast stream. Can not do the sort of on-demand streaming that Netflix would require.
I already acknowledged that torrents are good for some non-video applications. :)
I just can't ever see it being practical for video deployment. At least not in its current guise (thus whatever P2P model might succeed in the future wouldn't be authentically bittorrent).
If anything, with AWS and other "clouds" offering services for video delivery (Media Live, CloudFront supporting RMTP etc), I can see bittorrent becoming ever more irrelevant for video distribution. At least thats the trend Ive been observing recently. But tech can move in unpredictable directions so never say never.
Yes, torrents have some downsides - forever they can be worked through at some cost of quality. I might see some delays occasionally for example while I wait for an it is sequence chunk to arrive.
However for some content/use cases, perhaps that will be acceptable.
Right now the economics favor a dedicated and reliable changes from the source - e.g Netflix.
Perhaps one day another use case will appear where the massive cost benefits of piggybacking off other people's resources comes into its own.
I've done some work on with broadcasters distributing video assets over IP and honestly I can't see bittorrent ever being used. Once multicast over RMTP becomes reliable (which broadcasters are working towards) then the one advantage bittorrent has (reducing bandwidth from source) becomes moot for live services. Video on demand services would still have bandwidth issues but its less "spikey" and you'd factor that into the running cost of the service (and thus what you charge customers / advertisers).
Well, do you think that BT might have the potential to reduce bandwidth costs? And what difference might that make to the overall cost of doing business?
I think ISPs might quickly stop P2P video delivery if it became the norm because there's no way they would want to pay for the bandwidth. Or those costs will just get passed onto the consumers. Either way, it's not free bandwidth.
In terms of video streaming services, there's other ways to reduce bandwidth costs without resorting to P2P. Multicast being the prime candidate as that's where you send one IP packet which targets multiple different endpoints (rather than unicast which is the typical point to point method of IP comms)
There is a simple way to fix that. You wouldn't want to force everyone to upload anyway (some are on metered connections etc.), so what you would do is offer a nominal discount for uploading, but if you choose it then you don't just host what you watch, you host whatever needs more hosting capacity. Then random third parties can't tell what you watch because what you host is random.
This reduces the efficiency somewhat, but it's easy to minimize that: Don't give someone a hundred movies to host, give them one movie and have them host it a hundred times (over the course of a year).
residential customers have really poor upload speeds so how many customers would you need to have to replace 1 caching server collocated at an ISP? short of having FTTH, most customers would have < 10Mb/s up. a dedicated server can have multiple 10Gb/s links. you'd be looking at hundreds or thousands of customers per caching box. the "nominal discount" you're offering would have to be tiny for it to make sense financially. so much so that most customers wouldn't even bother.
If the nominal discount was $1/month, most customers with an unmetered connection would take it because it's effectively free money. From the company's perspective it could even be revenue positive by raising the price $1 and offering it back as the discount.
You're also not considering the power law distribution. If nineteen customers have 1Mbps uploads and one has symmetric gigabit, the average is >50Mbps. And the discount could be per-byte -- set it at half whatever the cost of central servers would be.
Watch how quick those unmetered connections become metered once this takes off :P Companies doing the distribution value control, this would be quite difficult to administer. People put up with flakiness only because they're getting something they shouldn't be, or because it's free anyways. Once people start paying, it becomes a very different set of expectations.
Put on your product hat: as Netflix, you want to control every aspect of the experience to ensure its absolutely seamless. When you outsource distribution to a huge set of third parties that becomes much harder to achieve.
My solution to this for a video streaming service is to have a master network that you control that feeds the data needed to cover the next 30 seconds of video playback, with future segments coming in via torrents. If any segment is still downloading once it hits the 30 second threshold the connection is dropped and that segment is downloaded from your more than capable network.
Connections that are known to be slow can seed just the ends of files with faster customers seeding the entire file.
I imagine the maths on this has been done and the cost of bandwidth is much lower than the cost of engineering the solution though.
Update the protocol to securely track who uploaded what, to whom, and you've got a market just like private solar panel owners and the energy grid. Pay the 1mbps and gigabit civilians proportional to how much they share.
It's entertaining to expand the idea beyond bandwidth...what about CPU/gpu time on my powerful consumer desktop sitting idle/off all day? Or storage space? The computer would need to be on to seed... Would people pay to also use the idle compute capacity?
Security is a problem, but it could be used to segment customers and providers. (Big corporations bidding for AWS capacity, tiny personal sites hosted on a CDN of cheap consumer hardware).
>most customers with an unmetered connection would take it because it's effectively free money
not really, because have to pay for the electricity costs of leaving your computer on 24/7. assuming a modest 50W power consumption for a PC, that works out to 36KWh per month, or $3.6 at 10 cents /Kwh. That alone wipes away your $1/month you're saving. also, that's a third of the monthly plan cost, how much of a discount can netflix really offer here?
as for hard disk space, I'll grant it's "free" space (as in the consumer wasn't going to use it anyways), but running consumer drives 24/7 with frequent seeks can't be good for them.
There is no need to have the consumer PC running 24/7. Also there is no need to have the same uplink speed as downlink speed. Consumers on average just need to reach a seed ratio of 1, while downloading and consuming the content and then possibly doing other stuff on their devices until the content is deleted. Actually, if they only reach 0.3 on average, then still 30% of the server infrastructure is saved, which is already a lot.
I gave 200MB/s down and 10MB/s up. I suspect this is more normal.
Yes, it depends on country. But I also suspect the total upload bandwidth for typical users in typical locations simply isn't there to make the idea work at scale.
because if you're offering customers $1/month discount to use their computers as cache boxes, but dedicated 1 caching box is equivalent to 500 customers in terms of capacity, then if the amortized cost of 1 caching box is less than $500/month, it's cheaper to go with the dedicated cache box than to pay customers to run cache boxes.
I’m watching netflix on my chromecast or smart tv. There’s no storage to speak of. I’m guessing that only PCs would be suited for what you’re suggesting, and that very few people watch on PCs.
For a company like Netflix, the thing to do there is to make your own such device and then you can put whatever hardware you want in it.
I'm actually slightly surprised that nobody at Netflix has already done that. Netflix Steambox would be a thing people would pay money for, and then you don't have Google/Apple/Amazon/Comcast/etc. diverting your customers to their video services because your customers are using their devices.
Also notice that you don't need 100% of users to upload. Having 10% of users with a 5:1 ratio still cuts your bandwidth costs in half.
I don't think Netflix subscribers are as worried about privacy implications of P2P as Netflix is(not interested in sharing data). Netflix already has all history data of all subscribers and can decide to sell/use it as it seems fit thought policy changes. As subscribers, it's no different to have our data visible to one profit driven company vs to all subscribers. Moreover, having this data available publicly would of immensely help in creating transparency among people and bringing then closer.
World would be a lot better place if we starts becoming comfortable with sharing our real thoughts/preferences openly, and it would simplify transition to truly public-sprited internet platforms based on etherium, etc.
I was of the opinion that Facebook used P2P / torrent protocol.
>> According to an interview in 2012 with Chuck Rossi, a build engineer at Facebook, Facebook compiles into a 1.5 GB binary blob which is then distributed to the servers using a custom BitTorrent-based release system.
Above part is taken directly from Wikipedia article on Facebook - "Technical Aspects" section.
I think it should be possible to build something akin to Tor (or even use the Tor protocol directly) that proxies requests and responses through intermediary nodes, so that the sender doesn't know who's the receiver and vice versa.
But you're probably right in that this is a relevant problem, as IP addresses are considered PII (in Europe at least), so leaking them could be problematic. That said, as Netflix already has a contract with each of its subscribers it shouldn't be too difficult to modify it to accommodate that kind of data sharing.
I imagine the real problem is ensuring streaming quality in such an infrastructure, as it's very hard to predict when a node will go down (e.g. user turns off laptop) or massively degrade in bandwidth (e.g. user starts downloading other content), and there's not much that customers hate more than interruptions when streaming movies (as you can probably attest firsthand). So given all these constraints it might be cheaper just using a CDN. Also, the bandwidth costs per customer probably make up a negligible share of overall costs (licensing is probably the biggest part) so it might not be worth optimizing this at all. I'd estimate that when I used Netflix I streamed maybe 20-30 movies / episodes per month (sometimes more), which would total at about 50-100 GB max (?) given the right encoding. For a company like Netflix that distributes probably petabytes of data a single GB should cost significantly less than 1 cent, so my bandwidth cost to Netflix would be between 0.5-1 $ per month (I even think it's much less probably).
I believe your interlocutor is suggesting that the Tor protocol makes tradeoffs in its design that tend to make it slow. This is demonstrated by slowness in the largest Tor deployment to date.
Also, have a large and single-purpose Tor deployment controlled by a single provider would mean there would be no privacy from that provider and anyone able to send them subpoenas. I think that this weakens the privacy provided to the point where the decision to use Tor at all becomes questionable.
Sure but the goal of such a system is just to distribute content more efficiently so protecting the users from the provider of the system is not the goal, you just want to protect the users from each other, if you like.
In that case, I'm afraid I don't see any advantages over a traditional CDN. Certainly not any privacy advantages, almost certainly not any performance advantages, and probably not any cost advantages either.
I don't know the real reason but for myself, I actually uninstalled Spotify after I found out about the P2P feature. There was a noticeable lag when I would play online console games and had Spotify open on a PC or laptop. It was never really an advertised feature and it kind of pissed me off that they would do that without a clear notification. Took me a bit to troubleshoot as well. I'm sure it was in the user agreement but come on, who reads that.
Interesting. Could there be a more privacy-friendly protocol involving cryptography? The Zcash cryptocurrency managed to hide transactions with zero knowledge proofs for example. It also reminds me of some cryptocurrency who asks other nodes about a bloomfilter of data, allowing them to receive what they want but more stuff to hide the real request.
The problem with P2P is that you have to make what you have public. That’s fine for a game update where it’s all the same, but for movies there are huge privacy implications.