SQL Server 2012 :: Fast Data Loading With Partition Switching Strategy

I’m looking for clearity on partition switching. The idea is to use many BULK INSERT statements into table dbo.X_n in parallel and when BULK INSERT for table dbo.X_n is completed, switch dbo.X_n into dbo.bigdaddy. I think this is the fastest way to upload a couple hundred GB of data.

In learning about partition switching (in part) from The Data Loading Performance Guide under Partition SWITCH, I hear the instructions to say copy the main table exactly to become a target. But in that same step (#1), I read that we need to change the default file group of the target (dbo.X_n) from the default file group. Then it says I need to match indexes and lists the filegroup as something we need to match with the main table.

As an overview of the partition switching strategy, I think the whole point of BULK INSERT with partitioning is to have seperate files (in same group) to enable concurrent uploading where each table has its own file. Once the upload is completed to a table (dbo.X_n) then we do the partition switch into the main table (dbo.bigdaddy). The data we just uploaded doesn’t actually move, just the metadata for it.

“Don’t have the same filegroup on your target as the main table. You must have the same filegroup on your target as the main table.”