We are able to begin to see the development of systems using peer to see concepts. But there's a place where peer to see isn't (yet) broadly used: website hosting.
Several projects happen to be released, but there's no large solution which may permit customers to make use of and also to lead to some peer to see webhosting.
I do not mean not-open projects (like Google Website Hosting, designed to use Google ressources, not users'), but open projects, where each user lead towards the hosting from the global website hosting allowing its ressources (cpu, bandwith) be accessible.
I'm able to think about several assets of these systems:
- automatic load balancing
- better locality
- no storage limits
So, why this type of product is not broadly used ?
EDIT: I believe the "97.2%, plz seed!!" problem happens because all customers don't seed all of the files. But when something where all customers equally lead to any or all this content is made, this issue doesn't happen any longer. Peer to see storage systems (like Wuala) are reliable, because of that.
The issue of proprietary code is applicable, too to the fact that an user may not know which content (possibly "bad") he's hosting. Interesting solutions.
I add one other issue: the latency wich might be greater compared to a devoted server.
EDIT 2: The discretion of code and data could be accomplished by file encryption. For instance, with Wuala, all files are encoded, and i believe there's no known security breach within this system (however i may be wrong).
So seeders wouldn't have numerous benefits, or couple of. However it would prevent people from beeing dependent of website hosting companies. And the like a decentralized method to host websites is closer from the original concept of the web, i believe.
This is exactly what Freenet essentially is,
Freenet is free of charge software which allows you publish and acquire information on the web without anxiety about censorship. To do this freedom, the network is entirely decentralized and marketers and customers of knowledge are anonymous. Without anonymity there can't ever be true freedom of speech, and without decentralization the network is going to be susceptible to attack.
Customers lead towards the network giving bandwidth and some of the hard disk (known as the "data store") for storing files. Unlike other peer-to-peer file discussing systems, Freenet doesn't allow the user control what's saved within the data store. Rather, files are stored or erased for the way popular they're, using the least popular being thrown away to create method for more recent or even more popular content. Files within the data store are encoded to lessen the probability of prosecution by persons wanting to censor Freenet content.
The greatest issue is it's slow. In transfer speed and (mainly) latency.. Even when you will get many individuals with decent upload throughput, it'll still do not be as quick a devoted servers or two.. The rate is okay for which Freenet is (posting data without anxiety about censorship), although not for hosting your site..
A larger issue is this content needs to be static files, which rules out it's use for most high-traffic websites.. For everyone dynamic data each peer would need to execute code (frightening), and would most likely need to retrieve data from the database (which may be another large delay, again due to the latency)
I believe "cloud computing" is all about as near to P2P web-hosting as we'll see for the moment..
Basically were to give a number of my Computers CPU and bandwidth with a p2p website hosting service, how could I make sure that it can't finish up getting used for everyone child porn or any other similarly disgusting content?
For the business I'm able to think about 2 reasons to not use peer hosting:
- Responsiveness. Peer located solutions are frequently reliable due to the huge quantity of shared assets, but they're also nutoriously unstable. Therefore the browsing experience is going to be intermittent.
- Proprietary data/code. If I have written custom logic for my website I'm not going everybody about the network getting access. Additionally you encounter privacy difficulties with customer data.
There's a IBM project known as YouServ relating to this argument: http://www2002.org/CDROM/refereed/84/index.html