If one turns on drive compression in Windows Server 2003 (ignoring dire warnings of performance), how does random access to that file work? Does Windows create a temporary uncompressed version of the file? Or does it uncompress it on the fly (yikes performance!). If the later, for a 5GB file, would it have to uncompress the entire file just to get at the last few bytes? Or is the compression algorithm such that large files are not compressed in one long stream but broken down into blocks?