Time Machine Question

its not going to save you a whole load, only a few hundred MB at most i would have thought, but someone may correct me, but caches arent that big. by the time you add incremental backups your not going to notice.

 

It would be less to start with ...but every hour it will add all the files that have changed. Meaning that original 30GB plus all the new files you have saved. It will continue to expand until your HD can't hold any more ..then it will start deleting the old stuff when it adds more new stuff.

 

Remember, though, that it's not going to save complete versions of each file every time the file changes. It's just going to save the oldest version it can, plus the differences contained in each subsequent version. In other words, just enough information to 'build' whatever version of the file it needs. I assume it's also going to use some sort of lossless compression wherever it can, but I also assume it's going to need your backup drive to be at least as big as the amount of space used on your primary drive.

 

This is not correct ...that is how it would work if it were using ZFS, but that's not available yet. It will copy the whole file under HFS+ ...but you are correct that it only copies what is "new" and not everything.

 

You may be right. I have to admit I don't know specifically how Time Machine is structured; it may save a complete copy of the most recent version, plus 'differences' to older versions, or even some other more complicated algorithm. I'm sure, though, that they're not going to save complete copies of every version of the file -- that would be unnecessary and take up way too much space. That's all I wanted to point out.

Even though ZFS isn't incorporated, I don't think that would prevent Apple from using a data recovery method like this that's built into OS X, and does the work 'by hand' in the background. I think that's completely within the capabilities of the system, if they want to progam it into the OS. It would be kind of like how spotlight keeps track of all the searchable data in all your files, and keeps that up to date 'on the fly' as you modify the files. In other words, I think it would be easier with ZFS, but still completely do-able without it.

 

Going on the information on the AppleInsider preview of Time Machine I would suggest that full copies of the files are taken for each new version. I can't see how doing diffs for each file would fit with their explanation of how it works.

Yes that takes a lot more space, but it also makes it faster. I'm quite happy to be proved wrong though.

 

I haven't seen the preview, so I'll check it out. I could be all wrong, but it just doesn't seem that hard to analyze and save the differences. The OS could do this in the background (again, similar to how Spotlight works). Downside with that method is that if one of the backup files gets corrupted it stands to take others (e.g. any earlier versions) with it, but if you store complete copies of each version they're independently salvageable.