Hello Everyone, I've been looking at doing a small scale test of an "embarrassingly parallel" algorithm as a proof of concept. I'd like to use hadoop & mapreduce so that I can scale it out once I've proven the concept. Because I have limited resources I'd like to do a community, points based reward system like seti@home or folding@home, but it means every participant would either need to install hadoop themselves or I would need to find a way to package it.
Everything I've found so far points towards the "download & install hadoop, then run your service on top of it." The problem I see with this is what I like to call Steve's law. "The number of people running your application is inversely proportional to the square of the steps required to install and run it." Thus I would like to keep this as simple as possible and just package it with the binary. Since I'm writing this in Java it seems I should just be able to import and package a jar file, but everything I've found says you need to setup a running daemon as well. Any thoughts? Thanks! /* PLUG: http://plug.org, #utah on irc.freenode.net Unsubscribe: http://plug.org/mailman/options/plug Don't fear the penguin. */
