Restart copy vs tape
Tape deserves respect. It is genuinely offline, cheap per terabyte, and it has outlived every technology announced as its replacement. If offline were the only requirement, tape would end the conversation.
Where tape is strong
Cost at scale. Longevity. True physical separation once the cartridge leaves the drive. For deep archive and regulatory retention, tape remains hard to argue with.
Where restore day hurts
An incident is a race, and tape restores are sequential. Locating the right cartridges, mounting, streaming, and praying the catalog server survived the attack, because without the catalog a tape library is a box of unlabeled boxes. Restore speed is measured in days for serious volumes. The process needs people who know the library, and those people are a shrinking guild. Rehearsals happen rarely for the same reason, which means the zero-errors digit in 3-2-1-1-0 stays unverified.
Where a restart copy differs
It optimizes for the first hours after total compromise, not for cost per terabyte. The critical core, held physically apart, restorable by the person on shift, at local disk speed, without a catalog server that may no longer exist. It answers a different question than tape: not "is our data somewhere safe" but "can we restart operations today."
The honest architecture
Tape for archive depth. A restart copy for the operational core. If your recovery plan's first 72 hours currently depend on a tape robot and one retiring specialist, run the five questions.