Hadoop on IBM Mainframe, what do you think?

I'd be surprised if this was interesting to anybody except those who already invested a lot in IBM z infrastructure.