2

Synology's Active Backup for Business has a feature for Destination Encryption which only seems to work with BTRFS. It works so that you have to set it up when creating the BTRFS storage. At this point you specify a password and when finished you are given an encryption key which you are told to backup safely. Going forward all backups to this location have Destination Encryption enabled and you cannot disable it. So it is also not possible to have backups without it on this destination. When you want to access or restore files from that destination you have to mount it which only works with the specified password or encryption key.

What makes me wonder is this:

  • When you reboot the Synology server it is not required to mount the destination and thus it is not required to enter the password or encryption key for backups to work (it would only be required if you/a user wants to read or restore from the location but seemingly is not required to write to that location)
  • When you add new clients the password or encryption key is not required anywhere (so not on the client to add and also not on the Synology server, regardless if the destination is currently 'mounted' or not)

So now I'm wondering, for this to be secure this would need to be an approach implemented with asymmetric cryptography. But since it seems to be quite performant (write speeds over 100MB/s even on my cheaper Synology test server are no problem) I'm not sure if it really is. Maybe BTRFS really has something like that? But I couldn't find anything regarding that (rather the opposite, it doesn't even seem to have native options for symmetric key based encryption)... Has Synology a custom BTRFS implementation in that regard?

Another way to make it at least somewhat secure would be the use of a TPM in any sort. But I'm having a hard time imagining an implementation with that and the behavior described above which is really secure...

On Synology's end I couldn't find much. There is a KB article which states that a 'unique machine key' is also used but it doesn't specify much more.

Can anybody clarify what kind of encryption might be at work here and if it might be secure? Or maybe it is easier to confirm that this cannot be fully secure? (I'm asking this in the context of an ISO 27001 risk evaluation with the main risk being the the Synology server getting stolen in whole. We would assume at least a short loss of the power supply so a loaded key in the RAM would be no problem but any way to retrieve the key/backup data after booting the Synology server without knowing/entering the password or key would be a problem.)

Improve this question
4
  • Do they have a whitepaper?
    – forest
    Dec 11, 2022 at 0:37
  • at least not at their 'usual place': kb.synology.com/en-us/…
    – Jey DWork
    Dec 11, 2022 at 0:56
  • My MacBooks ssd access speed is unchanged with and without encryption. That’s over 3,000 Megabyte per second.
    – gnasher729
    Feb 9 at 7:27
  • @gnasher729 your MacBook uses the T2 chip (which is Apples flavor of TPM) and thus the encryption is symmetric and expected to be only unnoticeable slower than non-encrypted. This is as secure as you trust the TPM and unfortunately highly irrelevant to my topic.
    – Jey DWork
    Feb 9 at 23:07

1 Answer 1

Reset to default
-1

From my point of view, the whole point of that encryption is preventing a third party accessing the backup on the target storage system, e.g. when you use the Synology Cloud offering or a S3 compatible target.

The Backup is encrypted on your side with the key and then uploaded to the target. But this encryption does not save you from a third-party obtaining the key from the system itself.

Improve this answer
3
  • Good for you. From the PoV of the ISO 27001 and therefore also auditors auditing its implementation the whole point of an ISMS in general is the highest possible level of security in proportion to the effort and protection required. Therefore regardless of how this encryption works the company has to know it and make a risk evaluation based on it. It is okay if it is not very secure but than it has to be known and documented so that a company knows in which case it might be good and in which they should better use a different approach. Just throwing an assumption like you doesn't help.
    – Jey DWork
    Feb 9 at 23:17
  • I don't see where I made an assumption. I just explained the use case behind the encryption. So in ISO 27001 terms, the asset that is protected is the external backup not the Backup on the machine. If you really want to know how this product works, you can always ask your Synology Partner.
    – Stefan Lorenz
    Feb 10 at 8:49
  • "But this encryption does not save you from a third-party obtaining the key from the system itself." Here you made the assumption that this is in fact a symmetric key because in case a public key from an asymmetric encryption method is used it actually would help. And you don't even state an indication that supports your assumption, much less any evidence.
    – Jey DWork
    Feb 11 at 1:09

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .