2

As you know, High Sierra (HS) includes a feature called “file cloning.” [EDITORIAL NOTE: Initially I'd written that, based on my discussion with the developer, I gathered Carbon Copy Cloner (CCC) can't back up cloned files. He's since clarified that CCC can't save cloned files as cloned files — it still does save files that have been cloned, but as complete files, rather than binary deltas. So all your files are properly backed up with CCC; it's just that the backup may take up more space than the original.]

Questions:

1) What actions do vs. don't create cloned files in HS? I did some of my own testing, and here's what I found:

I went to Finder-> Macintosh HD (in the sidebar, under Devices)-> Get Info. I then monitored the spaced used.

A) Duplicated two 3 GB data files in Finder (cmd-C then cmd-V). Here the space used didn't increase, indicating this procedure probably does create clones.

B) Opened a 128 MB data file in Excel. I then created a duplicate from within the program, using File->Save As. Here, the disk usage did increase correspondingly. [I did try using a larger data file, but it was taking too long.]

C) Opened a 500 MB data file in TextEdit. I then created a duplicate from within the program, using File->Duplicate. Here, the disk usage did increase correspondingly. [I did try using a larger data file, but it was taking too long.] 

Preliminary conclusion: Creating a duplicate within Finder results in a binary delta. Creating a duplicate within applications, both Apple and non-Apple, results in a separate stand-alone file (and takes much longer than creating a duplicate in Finder).

2) In addition to using CCC, I create parallel backups using Time Machine (though I prefer the former). I’m curious if the cloning function is also used to make Time Machine more efficient for non-cloned files. What I mean is this: In older OS’s, if a file had changed, Time Machine had to save the entire updated file. Thus, if one was using Entourage, where all emails were stored as a single multi-GB file, one would quickly fill up a Time Machine partition, since Time Machine would have to store a new multi-GB file with each backup. What I’m wondering is if, with High Sierra, Time Machine now only stores deltas of each backed-up file (i.e., just what’s changed). [EDIT: It appears this part has been answered in the discussion section, below: APFS doesn't change Time Machine's behavior, which was always to create binary deltas when the file configuration allowed it.]

  • 1
    Time Machine is currently APFS-unaware. It treats every file on the filesystem traditionally, and has always used binary deltas for in-place file updates. You should see no difference in behavior when it's sourcing from an APFS disk vs an HFS+ disk from previous systems. – vykor Oct 5 '17 at 22:16
  • @vykor: thanks. I looked into this further (your term-of-art, "binary delta", was the key to doing an informative Google search) and was able to find that Time Machine does use binary deltas to back up database files, but only when that file is stored as a sparse array. Since Entourage didn't do this, each TM backup had to save the entire file. See: hints.macworld.com/article.php?story=20091014152142736 and discussions.apple.com/thread/5194974?tstart=0 From what you've said, this situation is unchanged with APFS. – theorist Oct 6 '17 at 18:42
  • I second this question. I like to know which other operations, besides using copying in Finder, can create cloned files on APFS. For instance, which command line tools do that? – Thomas Tempelmann May 6 at 13:10
  • I would not use this terminology "create clones." I think of cloning as making a copy of something. A copy isn't made, but only a marker is made to the original file. It's something like an alias. Data get written to the storage device when data is written to the new file.. – historystamp May 6 at 23:18

You must log in to answer this question.

Browse other questions tagged .