Hi Morten, I'm hoping the reader of data does not have to be aware of the chunking if the data writer has chosen good chunk sizes and shapes for anticipated access patterns. Ideally, a reader should be able to write programs that access data relatively efficiently even using default chunk cache sizes, but that may not always be possible. It looks like we will be adding interfaces to permit setting and getting of chunk cache sizes for each variable independently in version 4.0.1. Thanks for your ideas on chunking issues. --Russ Russ Rew UCAR Unidata Program address@hidden http://www.unidata.ucar.edu Ticket Details =================== Ticket ID: WWQ-664381 Department: Support netCDF Priority: High Status: Closed