Jeff Hirsch wrote:
1) It sounds like you are saying that TM makes a complete backup every time it runs and requires a drive capable of holding multiple FULL copies of the source data.
My understanding is that Time Machine makes an initial backup of everything you ask it to cover and all subsequent backups after the first one are incremental and only contain the deltas from the original backup. Any differences.
(Is that no longer the case? If that were true, you'd need a massive TM drive capable of many many full copies of your data. That's never been the case as far as I understand.)
Time Machine uses a clever technical "trick" to implement incremental backups. The incremental part is an implementation detail. As far as the outside world is concerned, each backup is a complete copy of all files on all source drives.
If you ever attempt something like manually copying directories out of a Time Machine backup, or performing some kind of disk check, then you could run into an erroneous situation where you need an infinitely large hard drive or your file copy is expected to take 17 years.
A few years ago, Apple switched the Time Machine implementation from the old HFS+ format to the newer APFS format. That switch also necessitated changing the "trick" that makes Time Machine work. In the previous system, Time Machine used a directory hard link, which technically didn't exist. Modern Time Machine uses APFS snapshots, which are radically different, but more reliable, and less likely to cause the problems mentioned above.
You ask: "How does TM know what's changed between backups?" That's literally its job! You can see this clearly in the dialog where TM is "Calculating changes".
Don't read too much into what you read on dialog boxes. For any kind of long running process, any UI should be considered for user entertainment purposes only. There is no way to tell what is actually happening underneath.
It knows exactly which files are completely new, which are modified, and which have been removed.
But how? It literally has to check. That requires a disk access on the source and then another on the target. And then if it detects a change, it has to copy a file.
The older version of Time Machine used directories to shortcut the process and speed up the backup. The new design uses an APFS snapshot. That new design is going to result in different behaviour.
Each incremental backup (after the initial) is made up of the changes.
Once the internals are done, there are no more incremental backups. Each backup is a complete copy of the drive. It uses low-level APFS features like snapshots and copy-on-write data blocks to implement have a complete copy while reusing any data blocks that haven't changed.
Can you direct me to some documentation indicating that this isn't how TM works?
As I said before, this is an implementation detail. Apple doesn't document implementation details.
Perhaps things changed with one of the last couple of OS releases? I dunno. But this is literally the only user I support who has the problem. Everybody else uses a similar backup strategy and none of them are given this insane figure for estimated size needed to backup.
Perhaps there is something unique about this user's setup. Maybe one of the drives being backed up isn't APFS.
2) Your assumption about using an "older Time Machine drive" is erroneous. The user has tried brand new "out of the box" drives (and more than one!)
I think you misunderstood. If you have an existing Time Machine backup drive formatted in the older HFS+ system, then a modern computer will be able to continue to write new data to the backup. But users can no longer create those HFS+ Time Machine volumes. All newly created Time Machine volumes must be APFS. Another possibility is that this particular user is the only one with a new Time Machine drive. Any existing Time Machine drives will continue to work the same way they always have.
3) We may have different definitions for "constantly changing" in regards to the 6TB data I want to keep backed up for this user. The changes are typically to a fraction of the 6TB and not "the whole thing". So it may just be a matter of definitions/nomenclature. She isn't altering all 6TB regularly. Just adding, editing, or removing a minute fraction of that data on any given day.
So why back it up with Time Machine? It's like an accountant having client data for the past 20 years. Only the past couple of years is likely to change. So why spend the time, effort, and money on backing up all of the data that isn't changing?
Time Machine is a backup. It's not an archive. If there is data that isn't regularly being updated, then it needs to be moved to an archive. Otherwise, Time Machine may delete the data eventually.
