Peer to Peer technologies are getting more and more press lately. Verisign and Adobe signed a deal, The Venice Project, BIttorrent.com is announcing content deals, and of course there is the ever present treasure trove of illegal content available online via torrents.
The reason there has been excitement about P2P technologies built around BitTorrent type technology is simple. It saves bandwidth on file distribution and it creates the opportunity to speed the delivery of files, large or small. If it were able to live up to the hype, the notion is that how multimedia is distributed on the net, and its economics would change.
I'm not as sure it will as some others are.
The premise of the technically is to break up files into pieces and distribute those pieces on to the PCs of end users who have downloaded the BitTorrent type client. Then when a user requests the file to be delivered or streamed to them, rather than having to go to a host server, a tracker determines where all the file pieces are, and defines how the user reassembles them into a copy of the original on his or her computer as a file or a stream.
Thats the very, very simplific explanation of how it works.
From a business perspective, the important element is that if X number of people request a 1gbs file, rather than a host computer having to deliver files consuming Xgbs, the file is tracked among the peers and delivered using their bandwidth and resources , relieving the host of the bandwidth cost and obligation and hopefully speeding the delivery of the content
All good, right ?
For people creating content. Absolutely. For the end user, not so much.
P2P technology expects the end user to contribute bandwidth, hard drive storage and processing power. Something that in most cases, we all have available to spare. Most of the bit torrent client softwares have a "give to get " algorithm , In other words, it will opportunistically deliver content to you as quickly as the bandwidth you make available to it. So if you make 100k of bandwidht available to upload file segments being hosted on your PC, it will allow you to download up to 100k (there are other variables involved, but this is the simple way to understand).
All of this works very, very well in controlled environments. It also works well on a public internet tests when there are a lot of clients fully participating. In other words, they are offering their bandwidth, and open to seeding all content.
In real world execution however, it doesnt happen that way. There are multiple problems with P2P systems that could kill the golden goose.
1. Conflicting Clients . There are a ton of clients, with the number growing all the time. Although they work on basically the same source code and protocols, they all install and operate as if they had exclusive access. They want to control the PC so that they are in charge of what resources are available. When multiple clients are installed on a PC, not only does that create confusion among users, its a "last installed, first in charge" approach. THat approach and lack of respect for other clients will lead to user configuration problems. Thats not going to work. At some point they get considered to be malware and the clients will get uninstalled
2. End Users dont understand how P2P works, and once they do, they get concerned about giving up bandwidth. Most users dont know how to go in and edit the default settings. So even if they settle on a single client and are happy with just the content available on that network or to that client, they arent going to be happy about their banwidth being in constant use to save a content provider money .
3. The P2P model of seeding is a HUGE problem for those using wireless broadband with bandwidth constraints or per bit or per minute costs. People are going to wake up and find that they owe Verizon, Sprint, whoever a lot more than they ever thought possible because they installed a client on their Laptops. That could lead to these networks blocking the protocol.
4. There is a misconception that there is bandwidth savings for the end user. If you want to download a 1gb size file, 1gb of data will be delivered to your PC. There is no savings of bandwidth on the client side. In fact, the client is charged a bandwidth premium because after they have received the entire file, they are asked to particpate in the peering by delivering parts of the file to other users.
This in turn becomes an issue to services providers, whether DSL, cable, whoever. If quite a few users on a network segment are seeding files, it can slow down the network segment..
Its interesting to note that some feel that more than 55pct of internet bandwidth is consumed by Torrents. I dont know what percentage of internet users are using bitTorrent clients to acquire content, but it has to be relatively small. If that percentage doubles, what happens to performance on the net ?
If bittorrent client installation doubles or triples, does the pct of internet bandwidth used by torrents go from 55pct to 100pct ? Of course it wont work that way, but the 55pct to current client base ratio raises some very interesting questions about whether torrents truly do save bandwidth and can speed delivery of content
In conclusion, P2P is a product that tests great. In application however, it has a ton of challenges
A Question about P2P Technologies
Recent Posts
- Beating Google ? (5/14/2008)
Reader Comments
(Page 1)2. A few years ago in college P2P was a part of everyday life. We were always downloading music and video, eventually everyone figured out how to turn off the sharing with others aspect and it suddenly became alot more difficult to download anything.
In principle its a great idea, albiet illegal.
Posted at 5:53PM on Jan 11th 2007 by Michael Landers
4. I'm currently working with a company that addresses some of your concerns...
There's nothing you can do about #3 (wireless minute charges), but if you could centrally control that client software to manage the bandwidth used..
And as for the benefit to the client...what if P2P was the only way to access premium content? For instance, I don't forsee HD quality content being streamed using current technology...but what if you could stream live HD content only using P2P? There's value to the consumer for that, and they may willingly sacrifice some of their bandwidth (of which they don't pay more for) if they can get quality live video streams.
Of course, you bring up good points as you always do...but I see light at the end of the tunnel with the technology I've been playing with. It's not released to the public yet, but if you want to demo it, ping me..
Posted at 6:16PM on Jan 11th 2007 by Patrick
5. All good points, the key problem for project venice is how are they planning to stream video without buffering atleast 5 minutes for any potential peer outages.
P2P is not a solution for video files, especially when you're talking High-Def, or even DVD-quality. The best chance is video around the 200kbps range and the quality of that is shocking. Most people don't have symmetric connections, infact most upload connections are capped at around 256kbps and under T&Cs using your connection to host a commercial service is barred on most networks.
P2P is just not workable, it's a poor stopgap provision and someone will pay for it. If you think about tier-1 providers who have few home user clients, what happens when their networks start getting swamped with P2P traffic? They start charging ISPs more for their peering agreements, ISPs in turn pass their fees onto the consumer.
P2P is not a competitor of VOD because it can't provide the bandwidth for better video quality than you could get from flash.
Perhaps, another new technology buzzword is overhyped...
Posted at 6:18PM on Jan 11th 2007 by Adam Cains
6. Um... hello? Have you ever used BitTorrent on the public internet? It works GREAT! That's why it has risen to be such a high percentage of internet traffic. Maybe in the future excessive BitTorrent bandwidth will break the internet, but that would be due to the huge success of BitTorrent. So I don't really see any point to what you're saying here. Here's my reaction to your 4 points...
1. Conflicting Clients.
- All the BitTorrent clients I have tried work on all torrents. It's the same situation as music apps and MP3, and not really a problem. Most users will only ever have 1 BT program anyway.
2. End Users dont understand how P2P works.
- And they don't care, as long as they get their downloads.
3. The P2P model of seeding is a HUGE problem for those using wireless broadband with bandwidth constraints.
- But it's not a problem for the vast majority of internet users with unlimited wired broadband.
4. There is a misconception that there is bandwidth savings for the end user.
- Users don't care about the bandwidth, the big win of BitTorrent is it allows users to get their giant file downloaded in a lot less _time_.
7. Then there is the question of your own wait time and reliability, to get the file you need, or deliver it. When you talk about "files" in general and "people" getting files, you have to consider that there is a wide range of different purposes and priorities in delivering and receiving a file, or a bunch of files. For those who either have to or want to receive files via P2P, they risk never getting the file they want or waiting days to get it. With P2P, you are dependent on a hodgepodge of other peers who may be online, who may have decent bandwidth, and who may have their sharing bandwidth set to an "acceptable" bitrate. In my experience, even if 100 people are sharing the exact file you want, it still trickles in at rates that make a dial-up modem look fast.
The most practical, efficient way to send and receive files is through a service that is being rewarded for its performance and reliability in serving up files. Services like filesanywhere.com will deliver your files to receivers reliably, at maximum download rates, and they even send you an email when receivers access the file. Plus you can setup "protected" shares so that only the people you want to receive the files have access to them.
Again, it depends on the purpose and priority of the file sharing. For those who want to waste precious time TRYING to get a file, use P2P. For those like myself and my peers, who have very little time to waste, using a professional file storage service for 5 bucks a month makes way more sense than free P2P. For video, music, and photos, most of these services still give downloaders the choice to save the file locally or stream it and enjoy it online without downloading the whole file.
Posted at 8:56PM on Jan 11th 2007 by Meredith
8. I see the importance of this from a technology perspective, but how big of an impact (or how much of an impact) does this have on the typical home computer user? I don't think all that much, since they aren't often downloading huge files. I guess it's more of a business issue.
9. P2P is superior distribution technology in terms of efficiency - an intelligent implementation will result in less overall bandwidth usage because the content will be delivered from *servers* that are closer to the user. So the packets will therefore traverse less hops to get to the user, and, as you stated Mark, it is the same amount of packets regardless (ignoring potential compression schemes). However, superior user experience will beat superior distribution technology every time. So I think the success of P2P as a distribution technology to end users (it can also be used within the networks of content providers) is much more dependent on how the applications that use P2P evolve their front-ends. For long form video for example, will we get a better overall user experience than we do with NetFlix, broadcast, VOD, or (the new) Blockbuster?
Posted at 10:00PM on Jan 11th 2007 by Galeal Zino
10. The biggest problem with your post is that most of the users of torrents you speak of don't agree with any of that. Dustin makes a few good points.
People DON'T CARE if they give up bandwidth, they get FREE stuff FASTER. It's a trade-off any file sharer will take any day. There's no one who's dumb enough to use torrents where they pay per minute bandwidth charges... unless they're rich and don't care. Torrent users are much smarter than that...
The P2P clients won't conflict. The client with the best product, performance, reliability, and speed will rise to the top every time. Note reliability means they have to have dedicated uploaders and lots of users and lots of actual files. The client that contains all these attributes will win, it won't be the one who has the latest download available.
Most people who use torrents know how they work at least in basic form, much like how you described, Mark.
Posted at 10:19PM on Jan 11th 2007 by James Stevens
11. Well, after waving the red flag, you stand there in your bull-fighter's clothes and we don't know who you really are. And, on top of that, you're usually a lot smarter than this.
P2P file sharing has the potential to dramatically reduce the Internet backbone traffic moving the same bits around. If I can find them "locally", I can get them over a LAN, or a local sub-net. Otherwise, I have to go to Mecca for them, wherever and whatever Mecca is. P2P file sharing is a solution to a number of problems ... done at all correctly. None of the "let's share the latest pop tune or ripped DVD" projects started from technical first principles. Others have. The results were spectacular but you'd never know it.
Posted at 10:21PM on Jan 11th 2007 by Marty Heyman
12. I agree with James Stevens. Giving up bandwidth isn't a problem if in exchange you get a benefit from it.
It will be interesting to see how they try to do a business model of torrents. I just hope they don't "Napster" them.
Posted at 12:18AM on Jan 12th 2007 by Lamarr Wilson
13. I guess this will play a bigger role as broadband for computers starts to include tons of movie downloads, etc. I'd bet the average user doesn't require huge throughput right now.
Posted at 12:30AM on Jan 12th 2007 by basketball
14. When did you go from being a Maverick to being a Cranky Old Stormcrow? Seriously, lately you just seem intent on picking apart any technology you don't have a VenCap interest in.
1. Conflicting Clients: Have you actually used BT? Anything using the base BT code talks to all the other BT software. They all share with each other. Where's the conflict?
2. End Users don't know much of anything. That's why the GUI was invented - so that the cro-magnon majority of the consumer populace can click on the pictographs on the cave wall and grunt "Ooga-booga-unga-bunga" while they play video games and surf for pron.
3. Wireless Network costs are dropping faster than you can say "Google's Free Wi-Fi Network." Have you read a newspaper lately Mark? Or are you spending all your time watching B-Ball?
4. Bandwidth concerns: Vastly over-rated on many fronts. First off, as several others have already pointed out, for the end user, there really is no concern. As long as they get the content they want, nobody really cares where there excess bandwidth goes. For the Tier 1 and Tier 2 providers: Give me a break. There's so much dark fiber in the ground that Google (among others) are buying up vast quantities on the cheap now, knowing full well it won't increase in value for at least the next 10 years. Yes, Dorothy, there really is that much excess capacity in the ground. Given that there's enough fiber in the ground to handle the increased traffic for at least the next 10 years, the only real pressure on the network is in the routers and other hardware. Moore's law is keeping that technology well ahead of the bandwidth usage curve. Do your research and due diligence on the subject, Mark, that's how you became rich in the first place.
Sorry if I come off as shrill and harsh, but you're way smarter than that post makes you look.
Posted at 1:42AM on Jan 12th 2007 by John B. (No, really, my name IS that dull)
15. 1. You must know that all BitTorrent clients are compatible with each other so I guess you refer to the occasional news we hear about a BT client being biased to download more with less upload, and gets banned by some trackers or clients for doing so. This is what happens: unfair or otherwise malevolent BitTorrent clients get banned from the community of clients/trackers. This argument works against your point because I actually shows how this community is self-regulating efficiently.
The problem of running multiple BitTorrent clients on the same PC is real. The solution is to run BitTorrent as a kind of OS service, shared by all the applications that need to add torrent seeding or downloads to the queue.
2 & 3. As the other commenters I think upload bandwidth is not any issue for anyone. Except of course for the unlucky ones who are stuck with a capped connection. Of course for them P2P is not a solution. But I'm pretty sure these kinds of connections will disappear in the long run.
4. Bandwidth is essentially saved on the server's side, you're right about that. Even more importantly, LOAD (CPU, RAM...) is saved on the server side. That is a key point to understand what problem P2P solves that will still be relevant in the future.
Take for instance a time in the future when everyone is pulling HD content, say 15GB a movie, thanks to their 100Mbps fiber optics Internet access. How big a server do you need to serve millions of people pulling that kind of bandwidth? A very very big one, even considering the evolution of server powers, RAM, disk access and all,it would need to be a monster server farm.
I think, contrary to your idea, that it's not a problem if BitTorrent usage grows, and even triples, because the network capacities are growing too.
It all boils down to the evolution of technology, and Moore's law which is still relevant. But you have to understand that performance evolves thanks to progress in Networks, in Hardwares AND in Software. And P2P is the paradigm shift that we need to make the next generation Internet distribution happen.
Posted at 5:19AM on Jan 12th 2007 by Louis Choquel
16. I downloaded Pans Labyrinth at least two weeks before its US release. Plugged my laptop into my 32 inch LCD screen( that doubles as a tv via hdmi and my cable box) and hooked the cound up to my surround system and watched a fabulous movie before half of the USA even heard about it. Of course this was done using bit torrent, which I love. The content is there, taking advantage of it is the key. TV show, movies, music, applications its all there. I agree with the responses that say people dont care about bandwidth just yet. Getting a movie or music for free still has quite an appeal to it.
17. Mark,
Most of your posts show more insight than this one. Luckily your readers have broader knowledge than any individual can. The collective coments have corrected most of your mistakes(FUD?).
BitTorrent is a breakthrough in bandwidth efficiency. If BitTorrent is going to clog the internet consider the same amount of data being transfered via streaming or conventional downloads. Since the data trafic IS increasing rapidly the most effective technology to handle it will win.
How many years and billions of dollars have gone into streaming? Where is the mass adoption of the resulting technology? Streaming has major problems scaling. BitTorrent improves as the scale increases.
Only those frightened by content piracy are stil in denial of these facts. A TiVo type device running BT could solve almost all concerns, by only allowing the end user access to a simple search and play interface.
Posted at 8:00AM on Jan 12th 2007 by TiVolater
18. I don't think there are long-term issues in your list.
These are all valid points for the short term, and at any rate, it ignores the essential issue of IP.
Studios, distributors, publishers, whatever, are better off with streaming solutions, not for download solutions. This is the only safe way to distribute HD content on a broad basis. The cable companies' On-Demand model is superior to bit-torrent for most markets. Every other media has gone hosted, if you consider the difference between a website compared to a newspaper or magazine.
I was surprised when Vista shipped without a central Download app like Itunes or Steam, in part because of the brilliance of bit-torrent approaches. But like the media companies, I think MS is moving more to hosting content. Their MSDN Labs show that they could just host Windows remotely and charge people like a utility company, which is no doubt a long-term goal of theirs. They could partner with cable companies to distribute a locked Windows server in neighborhoods/basements, this server would then be accessed from thin clients in multiple houses (people's demands are really basic outside of gaming, which is moving to consoles anyway). Given the permanent issues of piracy and malware, etc. I think streaming media and hosted apps are the best long term bets.
My own company is moving this way due to outsourcing and the number of folks off-site, it makes too much sense to go away.
Posted at 8:30AM on Jan 12th 2007 by solomonrex
20. The terms of service for most consumer internet connections forbids hosting of any service. In spite of the name (Peer to Peer), participating in a P2P service means that I am hosting a service.
Then, of course, as others have hinted above, most consumer connections are unbalanced, so if I have a "fair" torrent client I am limiting my download speed to match my upload speed. I don't download pirated movies or music so maybe that skews my experience, but most of my downloads (FOSS) are faster than my upload speed, often much faster.
Why would I want to get slower downloads and violate the terms I agreed to when I signed up for my internet connection?
Posted at 9:57AM on Jan 12th 2007 by William Reeder
Add your comments
Please keep your comments relevant to this blog entry: inappropriate or purely promotional comments may be removed. Email addresses are never displayed, but they are required to confirm your comments. To create a live link, simply type the URL (including http://) or email address and we will make it a live link for you. You can put up to 3 URLs in your comments. Line breaks and paragraphs are automatically converted — no need to use <p> or <br> tags.
1. Great post - the problem of Torrent traffic can be a huge nightmare for participants up and down the chain.
Rough numbers based on your 55% - If Torrent Traffic usage triples (and other traffic stays stable) then Torrent Traffic would be nearly 80% of the Internet and total Internet Traffic would more than double.
Posted at 5:07PM on Jan 11th 2007 by Matt McLaughlin