Currently, we've got no terasort examples in flink, that's a pity!
We are now testing flink batch processing, but due to there are no offical implementation, so we have to find some projects implemented by third-party, but there are still a long way to go before really run it successfully(caused by bugs etc.)
Because terasort is a standard of benchmark testing for batch processing, it's necessary that flink should provide the offical implementation.
I made some changes on project https://github.com/ehiggs/spark-terasort
to implement the FlinkTeraSort, I think maybe I can put an PR for it and make it the offical implementaion for flink.