Hacker News Reader

Comment by namank | Hacker News Reader

Like http://en.wikipedia.org/wiki/Berkeley_Open_Infrastructure_fo...?

Right, but the idea is to outsource the computing power to clients for a given amount of money. Users are then paid small amounts of money for returned computing power from their inactive home PC. This would give more incentive for users to utilize these services, instead of the mere 527,880 active computers for BOINC. This would also provide a more affordable solution to owning/maintaining data centers.

I notice that the BOINC wiki mentions that users are attributed credits. "BOINC Credit System helps volunteers keep track of how much CPU time they have donated to various distributed computing projects." -http://en.wikipedia.org/wiki/BOINC_Credit_System

Right. Its a very interesting technical project but I`m at a loss for who your customers would be. I suppose the same people that use BOINC...? Because most companies that would need that level of power already have the infrastructure (then there is security...)

This is one market that would require some research. Again though, very attractive technically...you just need to find people willing to pay.

Could you build a reliable, fast enough web server on this? So that your master receives a request and hands it off to a node in the network which parses and responds to the client? THAT would work and could actually compete with Google Apps and AWS without their costs!

Yeah, there's a lot of research required for a project like this, so much so that I don't think one person could cover all the bases. Another reason I turned to HN.

Like you mentioned before, most of the bigger companies that need this level of power already have it. But the real question we need to ask is, do they want it? Does a cheaper price, less maintenance and greater amounts of power qualify for a transition to third-party cloud-computing? Or would security and the ability to fix any in-house issues out-weigh those?

As far as the server configuration, that's not my specific expertise, so that I'm not sure, but that sounds like a viable solution.

I think I'm going to write up an article on posterous about this, I'm interested to see how many people I can get interested in a project like this, in order to determine if I should dedicate any time to it.

I'll go into more details, such as specifications, features, issues and what the project hopes to solve. I'll post it as a "Show HN" when I get it finished, I'll also reply to to this comment so you know when it's up.

Also, sorry for the delayed response, in the middle of a flight to HK.

Reply on news.ycombinator.com