Mac OS X: Time Machine
Apple has added something they are calling Time Machine to the upcoming “Leopard” release of OS X.
My very first thought was that they’ve wrapped a gui front-end around filesystem versioning or snapshots. That would be the logical way to do this; otherwise you waste a tremendous amount of unnecessary space for backup:
How will Time Machine work for people who modify gigantic files on a regular basis? For example, consider a 1GB database file. If I make a tiny modification to that file, Time Machine writes out another 1GB file to record that change. As a result, drive space on my backup device could quickly disappear.
A while back, it was rumored that Apple might be working on ZFS for Leopard. Is that how they are doing this? Well, maybe, but others say “ZFS is not in the Leopard discussed at WWDC in any capacity”
Maybe they added versioning to HFS+?. It wouldn’t seem all that hard to do given the metadata capabilities already present. Versioning itself isn’t all that new; it goes back to DEC’s TENEX and VMS filesystems. Even SCO implemented undelete with versioning. But no, the descriptions of Time Machine don’t sound like versioning: they talk about having a separate drive for the backups. So that’s probably not it.
So do they really duplicate data over and over again? That seems so unlikely. At the very least a simple version control diff method would cut down on that tremendously. But we don’t yet know. Wikipedia’s reference says:
It is unclear at this time whether Time Machine should be considered as a simple back-up utility or as a complete filesystem-level version control mechanism. Apple’s website merely states that an API will be released so that third party developers can take advantage of Time Machine.
And that’s where it stands for now.
*Originally published at APLawrence.com
A.P. Lawrence provides SCO Unix and Linux consulting services http://www.pcunix.com