LTO Tape Backup: Where It Still Belongs
Tape is regularly described as obsolete and keeps not being obsolete. It survives because it does two things no disk-based system does as well, and both of them matter more now than they did a decade ago.
Why tape persists
It goes offline. A tape that has been written and removed from the drive is physically disconnected. Nothing on the network can reach it, encrypt it, or delete it.
That property was a minor convenience when the main threat was hardware failure. Against ransomware it is the point. Attacks target backup systems deliberately, and a backup reachable with production credentials can be encrypted alongside everything else. A tape in a cupboard cannot.
Cost per terabyte at rest. Once written, a tape costs nothing to store. Disk-based backup requires an array that consumes power, cooling and rack space continuously for as long as the data is retained. Across long retention periods the difference is substantial.
Those two properties are why tape holds its place in long-retention and compliance archives rather than being a legacy habit.
What tape is bad at
Being honest about this decides where it belongs.
Restore speed. Tape is sequential. Restoring a single file means the drive winds to its position, which takes considerably longer than a disk read. Restoring a handful of files scattered across several tapes is worse.
That makes tape a poor first line of recovery. If someone deletes a file at 10am and needs it by 11am, tape is not the answer.
Operational overhead. Tapes need handling, labelling, rotation and transport. Autoloaders reduce it but do not remove it, and off-site rotation is a manual process someone has to actually do.
Media and drive care. Tapes degrade, drives need cleaning, and both have environmental requirements.
The sensible architecture
Nearly every organisation using tape well uses it as one tier rather than the whole solution.
Disk for recent backups. Fast to write, fast to restore, covers the common case of recovering something from the last few weeks. Nearline drives are the correct tier, since backup is sequential and cost per terabyte governs — our SSD versus HDD guide covers why flash offers nothing here.
Tape for long retention and off-site. Older backups moved to tape, taken out of the building, and stored cheaply.
This satisfies the 3-2-1 rule cleanly: production data, disk backup, tape copy — two media types, one off site, and the tape copy offline.
Generations and compatibility
The practical constraint when buying, and it catches people planning a long archive.
LTO generations advance with each increasing capacity and speed. Compatibility follows a consistent pattern: a drive reads and writes its own generation, and typically reads back a limited number of prior generations, with write support extending back less far than read support.
The consequence for archives is significant. A tape written today will eventually outlive the drives that can read it. An archive retained for many years needs a migration plan — reading old tapes onto current media before the drives that can read them are gone.
This is the same problem as end-of-life hardware generally: the constraint is not when support ends but when the means to read your data stops being obtainable. Plan the migration while drives are still available.
Compression figures
Tape capacities are frequently quoted twice — a native figure and a compressed figure, with the compressed number based on an assumed ratio.
Actual ratios depend entirely on the data. Text and databases compress well. Already-compressed data — images, video, archives, encrypted files — compresses barely at all, and encrypted data essentially not.
Size against the native capacity. Treat compression as a benefit if it materialises rather than as capacity you have bought. Backup sets containing a lot of media or encrypted data will land close to native.
Encryption
A tape leaving the building is data leaving the building.
Modern drives support hardware encryption, which should be used for anything going off site. Two things follow.
Key management is now critical. An encrypted tape without its key is unrecoverable, which is the correct behaviour and a genuine risk. Keys must be backed up separately and must survive the loss of the systems that created them.
This is one of the most common backup failures in practice: the backup existed, the tape was fine, and the key was on the server that was destroyed. Our backup guide covers what else gets forgotten.
Disposal. Tapes at end of life hold data and need proper handling — physical destruction or documented sanitisation, as covered in our data sanitisation guide.
Testing
The same rule as all backup: an untested restore is an assumption.
Tape adds specific failure modes worth testing for. Media degradation, which is invisible until a read fails. Drive cleaning requirements, since a dirty drive writes tapes that read poorly later. Catalogue dependency — some systems need their index to locate data on a tape, and losing it while keeping the tapes can make restoration slow or impractical.
Test by restoring from tape periodically, including older tapes, and confirm the process works without the original backup server.
Common questions
Is tape backup obsolete?
No. It does two things disk does not: a written tape removed from the drive is physically offline and unreachable by ransomware, and once written it costs nothing to store where a disk array consumes power and space continuously.
Should tape be my only backup?
No. Tape is sequential, so restoring a single file means winding to its position and takes far longer than disk. Use disk for recent backups where restores are common, and tape for long retention and the off-site offline copy.
Will my old tapes still be readable in ten years?
Not necessarily. Drives read back a limited number of prior generations, with write support extending back less far than read support, so tapes eventually outlive the drives that can read them. Long archives need a migration plan while compatible drives are still obtainable.
Should I size against native or compressed capacity?
Native. Compressed figures assume a ratio that depends entirely on the data — text and databases compress well, while images, video, archives and encrypted files compress barely at all. Treat compression as a benefit if it materialises.
What is the biggest risk with encrypted tapes?
Losing the key. An encrypted tape without its key is unrecoverable, which is correct behaviour and a real risk. Keys must be backed up separately and survive the loss of the systems that created them — a common failure is the key sitting on the destroyed server.
Tell us your retention requirement and restore expectations and we will help size the disk and tape tiers together.
