We do this all the time but use the Vcenter datastore browser. I seems to do a copy of the full 20GB file looking at how long it takes but when the file is stored on the new datastore it is still thin provisioned.
Do keep in mind to copy the metadata file as well. When you look in the datastore browser every appstack has 2 files related to them. The actual VMDK flat file and the metadata file which is just a few kb in size.
When importing the only thing that is lost is the assignment. When we do this with writables it even import the status of the writable if it is enabled or disabled. Works like a charm..
Thanks. I did do it the way you do and sure enough worked great. And yep, it thin provisions the disk when placing on the destination datastore. Thanks for the reply and for your help!