4 ms·

The crypto in Borg is bad. Even has its own page in the docs about how bad it really is. The tl;dr is "there is one AES key per repository and we use this with

The crypto in Borg is bad. Even has its own page in the docs about how bad it really is. The tl;dr is "there is one AES key per repository and we use this with AES-CTR and trust the remote server to not fiddle with the counter". https://borgbackup.readthedocs.io/en/stable/internals/security.html https://borgbackup.readthedocs.io/en/stable/internals/securi... Last time I checked there was still no work being done on improving performance. Multithreading has been on the agenda for years and will probably never come, much like improved crypto.

m3nu 8y ago
Like many open source projects, Borg is low on developers. The two points you mention are both known and on the agenda for this year. Since you seem to have experience with crypto, why not add a PR? It's all in Python and fairly readable.
- dataflow 8y ago
  >> Multithreading has been on the agenda for years and will probably never come, much like improved crypto. > The two points you mention are both known and on the agenda for this year This is a rather strange response to the above...
- gpm 8y ago
  When someone says something is the "holy grail of backups" and it is "gpg-encrypted" I assume it does that well, not it does it poorly right now but might improve in the near future. Thus his post was valuable to me. "Why not add a PR" is not a reasonable response to something like this. I'm looking to use a backup system, not create one. If something works so poorly that I'd have to start submitting PRs before using it, I'm far more inclined to go elsewhere.
- blattimwind 8y ago
  Borg isn't really low on developers, it has high churn, i.e. repels developers.
are595 8y ago
I'm not sure that's a fair summary. It's very clearly stated that there is authentication performed after encryption. The MAC should prevent the remote server changing the counter or any encrypted bits.
- blattimwind 8y ago
  > When the above attack model is extended to include multiple clients independently updating the same repository, then Borg fails to provide confidentiality
  - 8fingerlouie 8y ago
    Which is why multiple clients are not supported within the same repository.
    - blattimwind 8y ago
      They absolutely are supported.

m3nu 8y ago

Like many open source projects, Borg is low on developers. The two points you mention are both known and on the agenda for this year. Since you seem to have experience with crypto, why not add a PR? It's all in Python and fairly readable.

dataflow 8y ago
>> Multithreading has been on the agenda for years and will probably never come, much like improved crypto. > The two points you mention are both known and on the agenda for this year This is a rather strange response to the above...
gpm 8y ago
When someone says something is the "holy grail of backups" and it is "gpg-encrypted" I assume it does that well, not it does it poorly right now but might improve in the near future. Thus his post was valuable to me. "Why not add a PR" is not a reasonable response to something like this. I'm looking to use a backup system, not create one. If something works so poorly that I'd have to start submitting PRs before using it, I'm far more inclined to go elsewhere.
blattimwind 8y ago
Borg isn't really low on developers, it has high churn, i.e. repels developers.

are595 8y ago

I'm not sure that's a fair summary. It's very clearly stated that there is authentication performed after encryption. The MAC should prevent the remote server changing the counter or any encrypted bits.

blattimwind 8y ago
> When the above attack model is extended to include multiple clients independently updating the same repository, then Borg fails to provide confidentiality
- 8fingerlouie 8y ago
  Which is why multiple clients are not supported within the same repository.
  - blattimwind 8y ago
    They absolutely are supported.