Amazon MapReduce

The book I read to research this post was Amazon MapReduce Developer Guide which is an excellent guide and which was a free download from kindle. The idea behind this service from Amazon is it works primarily with Elastic Compute Cloud and say if you run a website and it gets a lot of traffic at the weekend you can configure how many instances and other things like total memory are used. Say if you are a company you might have instances open that you aren't using ie a database at weekends and you can set rules for how this is managed. Often if it has to shut down instances it will either shut down the ones that have been open the longest or ones that haven't been configured within MapReduce. A frequent setting is to set that when the processor reaches 90% output a new instances is opened. Another frequent setting is it will try and keep your total down by trying not to have any extra instances open for more than an hour. You can use it with Cloudwatch and/or SNS or Simple Notification Service that let you know what is happening. Also if you are repairing it and don't want instances going off continously you can suspend it. You have to install java before you can use it which has to be at least version 5 which you can get from http://sun.com
Alternatively you can use the Amazon Java Development Kit. You will often use a command line and there is an API command line tool you can install on your computer. Also don't forget if you are using Elastic Compute Cloud you need to provide a Windows Server 2008 operating system.