In der Nachricht vom Thursday, 6 June 2024 23:43:56 CEST schrieb Steve Litt:
> It sounded good until the deduplication and compression. I view
> deduplication and compression as negative parity bits, such that even
> in a text file, one flipped bit messes up everything.

This is basically true but mostly irrelevant for several reasons:

- The software is designed to run on a dedicated backup server. On mine it
saves the backups on a RAID6 on four disks, so there is redundancy, it has ECC
RAM, UPS, asf. that should make the machine more reliable. For sure more
reliable than a single disks that gets carried around.

- This solution should not be and is not the only mean of backup I use. I run
Syncthing on my machine and another one with "simple file versioning" turned
on, to have a life copy and some older versions of my data. This is in another
location than the backup server running BackupPC. These files on disk are just
a normal copy, thus they are accessible without a special software. In
addition I do rarely a clone image of my laptop's with dd to a spare disk,
that stays offline, ready to replace the current disk physically.

- On the backup server there are time series going back almost four years.
There are 12 machines getting their backup pulled by this machine, which is in
total prior to pooling and compression 160TB. This gets reduced to 4.2TB with
BackupPC. I could not save such long time series and that many machines
without this technology. And yes, there was already a need to go back two or
three years and I was glad I could.

> Is there a way in BackupPC to disable deduplication and compression?
Compression can be turned off. I haven't seen anything to switch off
deduplication.


Even if there is a certain risk in the technology of deduplication and
compression, the question is how you design your backup concept as a whole to
handle that risk.

Regards, Adrian.


PS: A word to backing up DBs: I run a script to dump all databases to an SQL
file, the script uses lock table statements and gives a consistent state of
each DB like this. I exclude the running DB's from my backup. Similar can be
said about VMs.