The DKRZ scientists are using NetCDF files to store large amount of data. We have observed, that the amount of data, and consequently, the time spent for I/O is growing constantly over time. This is an undesireable tendecy. We are using different approaches to relieve the situation. One of them is optimization. To see the impact of our optimizations we need a precise benchmark tool. Unfortunately, we couldn't find a suitable one for NetCDF file format. This was the decisive impulse to start the development of NetCDF-bench.