I think that I have successfully partitioned a 125 GB database with the data in place but I would like confirmation that my thinking is correct. The partition field happens to be the first field in the primary key.
The steps were:-
1. Create partition file groups using ALTER DATABASE.
2. Create partition files using ALTER DATABSE.
3. Create the Partition Function.
4. Create the Partition Scheme to tie it all together.
5. Drop and recreate the index (primary key).
6. Use DBCC SHRINK.
7. Use DBCC REINDEX.
The database was stored on a USB 3 hard drive and steps 5 and 7 took 10 to 11 hours to complete.
At the completion of step 5, the partition file sizes indicated that data was stored in them but the original mdb was still very large. After executing SHRINK, the original mdb was much smaller in size which I assume this means that the job is not complete until SHRINK is executed.
I am not sure whether step 7 was really necessary. It is usual to do this after SHRINK but as I had just dropped in recreated the PK, perhaps it was not really necessary.
What I am looking for is confirmation the above procedure is correct for partitioning with data in place. Everything certainly appears to be OK.
R Campbell