Krishna Sapkota thinks this is interesting:

have Hadoop configuration files containing the connection settings for each cluster you run against and specify which one you are using when you run Hadoop applications or tools.