The Zombie Stack Exchanges That Just Won't Die
Assuming that preserved content is kept in archival storage on spinning disks and that standard backup procedures are in place, is one complete copy (separate from the backup) of the content enough, provided it is stored remotely?
Bill Lefurgy
The conservative answer would be three copies so you have a tie-breaker in case two copies disagree. If you have something like two online systems and a tape backup, that should qualify and the issue is particularly improved if you store fixities separately (also redundantly and checksummed, of course) and thus have a high degree of confidence in your ability to state whether any particular copy is correct.
The other issue, which might have been implied by your “standard backup procedures“ note, hat I wouldn't trust any copy which hasn't been recently verified: spinning disks avoid a significant class of problems but they're still prone to bitrot and too many filesystems will silently return corrupted data. Something like an object store or modern filesystem (ZFS, btrfs, ???) which performs active scrubbing against a separate fixity calculated from the original data is a much safer proposition than, say, a neglected departmental backup server.
To summarize: yes, if both active copies are on modern storage systems and you use a separate fixity checking system. No if you're forced to rely on legacy storage since you'd have to devote significant resources to auditing and validation and, presumably, rely on things like quorum tests.
Chris Adam's answer addresses problems with backup media, but there is also the issue of what is sometimes called "business continuity" - getting your organization back up and running in the event of a disaster.
When making business continuity plans, you have to carefully consider various scenarios and whether or not it is even possible for the organization to recover in a particular scenario.
So take a software development company. As long as most of the company's key personel survive, the company can recover if it has good, recent backups of its intellectual property (source code, customer data, contracts, etc). Assume that the company's main facilities have been completely destroyed and that all you have is whatever was off-site.
If all you have is a bunch of tapes that have to be retrieved and then loaded somewhere, getting up and running again is going to take a significant amount of time. If one copy of each tape is all you have, and that one tape has a problem, then the organization is going to lose precious data during a time when it can ill afford to lose anything.
The general rule that I go by is: three copies, one of them a local backup that is used for simple recoveries of accidentally deleted files, computer crashes and the like. The other two copies should both be offsite and in separate locations. Or use a cloud based backup service such as CrashPlan that takes care of the multiple redundant copies in multiple locations details for you.
A common mnemonic for this is 3-2-1:
As a rule of thumb, it's not bad.