Hi Michael, The advantage is for dynamic collections of data, e.g., archives of real-time data that just keep growing. It allows such aggregations to avoid using the latest dataset (which might not be complete) as the template for the aggregation. Hope that helps, Ethan Michael McDonald wrote: > Package Version: TDS 4.2 > Operating System: RHEL4 > Hardware: 4x Dual Core AMD Opteron(tm) Processor 880, 32GB RAM > > The default NcML Aggregation method between THREDDS v4.1 and v4.2 > changed from random to penultimate. > > Why was this done? > > Is there a distinct advantage to using penultimate over random for > aggregations? > > References: > http://www.unidata.ucar.edu/projects/THREDDS/tech/tds4.1/reference/ThreddsConfigXMLFile.html > > http://www.unidata.ucar.edu/projects/THREDDS/tech/tds4.2/reference/ThreddsConfigXMLFile.html Ticket Details =================== Ticket ID: TOA-282078 Department: Support THREDDS Priority: High Status: Closed