In the case of Windows Update, Windows would be making changes that it wouldn’t want the other programs to see, but those other programs would also be making changes (and potentially to the same directories, though hopefully not the same files). Remember that a shadow copy works on a whole volume, and divides it into “the contents of the volume at that moment” and “the changes since then”. We could certainly create a “reverse shadow copy” that works this way but there’s another problem. What Windows Update would presumably want to do is the reverse - it would want to make changes to the Windows directory, while everyone else sees an unchanged version of it from the moment before the update started. The shadow copy is unchanging, and normally only one program sees it. So immediately there’s a bit of a disconnect between how shadow copies work and what Windows Update needs to do.
By using a shadow copy, a backup program knows that it’s capturing a consistent view of the system at one moment in time (the moment when the backup started). If, say, you were installing a program while the backup was going on, the backup might contain only half of the program’s files, with all sorts of bad results.
A problem with Windows backups programs prior to the implementation of shadow copy was that the backup was sometimes inconsistent, because the backup contained the version of each file that happened to be present at the moment that the backup program got to it. Shadow copies are typically used for backup products. That’s because when one of these programs says “Give me the data in file X”, Windows starts by reading the shadow-copy version of file X, but then checks to see if any changes have been made if so, those changes get applied before the data is returned to the copy. All the other programs don’t typically know that the shadow copy exists, and they see the changes made since the shadow copy was created. Typically, the program that created the shadow copy is the only one that sees its unchanging contents. So in other words, the volume is essentially split into two parts: one that holds the contents of the file system at the moment the shadow copy was requested, and one that holds all the changes that have been made since then. While the shadow copy is active, writes to that hard drive don’t overwrite the existing contents of the volume instead, they’re redirected to a special part of the file system. A shadow copy essentially lets a process take a snapshot of a volume (C:, D:, etc.) at any given instant. In order to explain why shadow copies don’t really help us, let me first explain what a shadow copy is.
0 kommentar(er)
