This may be too advanced of a question to post in the Sandbox forum. At a high level you would need to set up eclipse and edit the build path to include the hadoop library. You could then build out your java code the leverage a jdbc connection to query and then subsequently process the records. This could be done on the sandbox by copying your resulting .jar to the sandbox VM and running it with the hadoop jar command. There are many sources of examples on the internet, this one may be useful to start with -> http://developer.yahoo.com/hadoop/tutorial/module3.html#eclipse. I hope this helps.