On 4/25/2011 11:30 AM, Roy Mendelssohn wrote:
Hi All:
We just converted one or our larger datasets (larger in terms of the number of
files that are aggregated) into compressed netCDF4. There is a substantial
savings in storage, but we wanted to do a series of tests to see what hit in
access time we would take, if any, wsince many of our users will make requests
involving a lot of time periods.
In order to design these tests properly, we need to get a better understanding
of how the TDS handles netcdf4 datasets that have compression. Are the
decompressed data cached, or more accurately cached any differently from data
read from an uncompressed series of netcdf3 files, or since the decompression
is handled automatically on the read, is everything handled the same after that?
We would also be interested other peoples experience with compressed netcdf4
files in TDS, in particular when the extracts are not synoptic, but cover a lot
of time periods in a region, or make a lot of very small calls to a large
number of time periods - such as we need to do for tagging data.
Thanks for any info,
-Roy
Hi Roy:
I assume you mean internally compressed, not externally (like zipping up
a file) ?
