Thank you for your post.
When first logging into a hosted database solution, a temporary folder is created on your machine. When you perform a sort, temporary data is captured and cached to your machine. Therefore, the initial sort will take longer. Once the data is cached to your machine, subsequent sorts will be faster as the information is now cached on your machine.
Is this why I see "copying records" when performing certain operations on a hosted file?
Well... there are a few things missing:
You got 50'000 records. Where do they come from? Related table? Calculation fields?
If so, how is the relationship built?
Network.. LAN, WAN, celluar? WLAN?
Was something installed in the meantime? Virus-protection, etc?
Was it OK before switching to Yosemite?
Over WAN, this kind of sort could take that much time - depending pn the structure
** Sorry for typing mistakes - but it's impossible to edit postings on iPad here
In answer to your questions.
50,000 records start at over 100,000 but slows at 50,000. They are records from another layout used in this layout.
Relationship I'm not 100% sure on.
Network is LAN, only 1 or 2 using Yosemite.
Nothing new installed and server not using Yosemite.
Its only recent but across all machines.
Do you need the data from all 100,000 records for your report or only a subset of that data?
If you only need to print a report from a much smaller subset of this data, there should be a way to modify your design so that sorting the entire set of records is not needed, only a sub set.
Efficient sorts in database systems are usually referred to as O * N * Log N algorithms. (inefficient ones are O * N * N) where N is the number of elements to be sorted. Note that the time is a function of the number of items to be sorted and that the time required increases geometrically (on a curve, not a straight line if you graphed time vs. number of items). Thus, you may not have seen this delay previously due the fact that you didn't have as many records to sort.