I have a pretty large aggregated, gridded dataset: 1222x1183 and over
27,000 time records. The data is aggregated along the time dimension.
What I can't understand is why it takes so long when I request variables
that do not have my aggregated variable (time) as a dimension, i.e. lat
and lon. I have taken 21 files from my dataset and put them in a
separate aggregated dataset and the response time for lat and lon is
less than a second compared to several minutes for the 27,000 time
records case. The time test were performed using ncks on a TDS 3.17.3.1
server.
I'm sure there's a reason why this isn't the case, but I would think
that when a non-record-dependent variable is requested from and
aggregated dataset the thredds server would just look at the first file
in the aggregation and thus be just as quick as a much smaller
aggregated dataset. Isn't it safe to assume that all variables that are
non-record-dependent have the same values throughout the entire
aggregated dataset? If not, what file is thredds reading the values
from? (first file in the aggregation? last? somewhere in the middle?)