Real bulk-loading is faster - for example with Sybase the bulk-load API will create the binary image of the row on the client and send this image to the server. This means that the server only has to allocate the space for the new row and write the row to the disk, but doesn't have to perform any other work (such as converting between data types, maintaining indexes, calling triggers, etc.)

The downside is that there is no consistent API for bulk-load calls, so portability is almost impossible to achieve.

Michael