Go-nuts*, wouldn't you like to get your complex data analysis programs to work on hadoop and hdfs?
We'll start small, trying to get Go to work on hadoop, then slowly try to skip the Map-Reduce part and run dist.go or go-circuit with just HDFS.
For prep, try to get hdfs.go working - https://github.com/zyxar/hdfs
And if there's time, let's try and get YARN to spin up go-circuit clusters for processing, with zookeeper to keep track of spun up nodes.
That is possibly a huge task, but what would help would be a thorough understanding of gocircuit.spawn() - http://www.gocircuit.org/spawn.html
* (yes, that's what they call people who like Go)