My team has asked me to evaluate the feasiblity of writing an I/O module capable of storing model results directly into a large Hadoop cluster (targeting Hive and/or HBase). I've downloaded all the I/O stack documentation and have started reading through. Before I spend too much time on this, can I ask a few quick questions?
- Has anyone written such a thing already?
- Can replacement I/O modules be written in C or C++, or is Fortran a requirement?
- I note that the default I/O module talks with netCDF libraries. Are these written in Fortan as well?
I will continue studying the documentation, but any input or thoughts would be welcome.