2012 Unidata NetCDF Workshop > Formats and Performance
14.2 Classic File Format Performance
Most netCDF classic file access is fast, but
a few kinds of access may be slow for large datasets.
Understanding the netCDF classic format makes clear why.
Slow operations on classic format data
- adding information to the header of a large file (may have to
move data to make space)
- adding new dimensions, variables, or attributes to an
existing file
- modifying attributes to take more space, for example longer text strings
- (rare) even renaming existing dimensions, variables, or attributes with
longer names!
- accessing data from record variables a variable at a time, rather than a record at a
time
- accessing cross-sections of data in a different order than they
were written, for example one value per disk access
2012 Unidata NetCDF Workshop > Formats and Performance
