2 ms·

Would be really interested to see a write-up of those poisoning variants. Presuming cold backups are tamperproof, it seems to me poisoning really only make sens

Would be really interested to see a write-up of those poisoning variants. Presuming cold backups are tamperproof, it seems to me poisoning really only make sense when fresh data is significantly more valuable than historical (pre infection) data? If you had a hypothetically secure recovery testing procedure, then it should pick up the data is corrupted right? In which case you could tune the frequency of when you run it to reduce the amount of potential fresh data loss.

koheripbal 5y ago
> only make sense when fresh data is significantly more valuable than historical This is the case for basically any transactional/accounting system. If you lose even one day's data, if often would make more sense to pay the ransom than try to reconstruct the transaction flow for that day (or portion of the day) that was lost.