When you copy files and directories to a new location (or restore them from a backup image), there will be very little possibility that the occupied space will be the same. Consider a directory that had 5 million files in it. All the files were removed so the directory is empty. HOWEVER, the space needed to remember all those file names in the directory is NOT reduce. There are a million free entries in the directory that can be used. When this directory is created at the new location, it has no entries and will be significantly smaller in terms of occupied space at the target.
Conversely, a 'sparse' file has very large pieces of undefined space between records. A program writes record 1 then lseeks to record 1000000000 and writes one more record. The file occupies very little space. But when it is copied, the file is read sequentially. Since there is no way to inform the copy program that these records are undefined, a stream of zeros (nulls) is supplied and not the destination copy of the file is one million times larger than the original file. Sparse files are very common in database programs.
So to verify that a copy is correct, you count the number of files and directories. And if you are very paranoid, then run a cksum on every file and compare the values with the original files. Note that cksum will produce identical values with a sparse file and the copy.
Bill Hassell, sysadmin
