LTO tape is not a replacement for your NAS. It’s not even a great fit for most people’s backup strategy below a certain scale. But at the point where disk-based cold storage starts costing real money every year just to sit there doing nothing, tape’s economics flip in a way that’s worth understanding before you dismiss it as a museum piece.

What LTO actually is

LTO (Linear Tape-Open) is an open standard, jointly developed by HP, IBM, and Quantum, specifically so tape media and drives wouldn’t be locked to one vendor the way earlier tape formats were. It’s now in its ninth generation, and each generation roughly doubles native capacity over the last:

Compressed figures assume a 2.5:1 compression ratio, which is the marketing number every vendor quotes. Real-world ratios depend entirely on your data. Media files (the bulk of most data hoarders’ storage) are already compressed and won’t compress further on tape, so treat native capacity as the honest number for planning and the compressed figure as a best case you probably won’t hit.

Tapes are written and read sequentially, like a cassette, not randomly like a disk. That’s the single biggest thing to internalize before buying one: LTO is an archival medium, not a working filesystem. You don’t browse it. You write a backup job to it and you restore from it. Trying to use it like a slow external drive is the fastest way to get frustrated with the whole format.

The cost math that makes people reconsider

This is the actual reason to care about a decades-old format in 2026. Compare the economics at real hoarder scale.

A single LTO-8 tape holds 12TB native and costs somewhere in the $40-70 range new (used/off-lease media is cheaper but riskier, see below). That’s roughly $3-6 per TB for the media itself. A comparable amount of cold storage on spinning disk, even shucked externals (see the hard drive shucking guide for that angle), runs closer to $12-18 per TB once you factor in the drive, and that drive is a moving mechanical part sitting powered-off in a closet degrading anyway, or powered-on and burning electricity to protect data nobody’s touched in two years.

The catch, and it’s a real one, is the drive itself. A tape drive with a SAS interface (LTO doesn’t do SATA) runs anywhere from a few hundred dollars used for an older generation to well over a thousand new for current-generation hardware, plus you’ll need a SAS HBA if your box doesn’t already have one (the same kind of card covered in the HBA and IT-mode flashing guide, though tape drives typically want a straightforward SAS initiator connection rather than an IT-mode-flashed card meant for disk pass-through). That drive cost is the whole reason tape only makes sense past a certain scale: it’s a large fixed cost amortized across however many tapes you eventually buy. Buy one tape and the true cost per TB is absurd. Buy forty tapes over a few years and the drive cost per TB approaches zero while the media cost per TB stays flat, which is the opposite of how disk economics work (every additional TB on disk costs roughly the same as the last one, with no fixed cost to amortize away).

Rough breakeven, in practice: if you’re archiving under 20-30TB total and don’t expect that to grow much, buy more disk. Past 50-100TB of genuinely cold data you’re comfortable not touching for months or years at a time, tape starts winning on both cost and long-term reliability.

Where it actually fits: a third tier, not a replacement

Tape is not your primary storage and it’s not even your primary backup. Think of it as a third tier that sits underneath both:

1. Primary array - your live, working data. ZFS or whatever pool you’re running day to day, with the redundancy covered in the RAID redundancy guide and the corruption detection covered in the bit rot guide.
2. Backup - a second copy, on different media, ideally off-site, per the 3-2-1 guide linked above. This is what you restore from on a normal bad day: a dead drive, a fat-fingered rm -rf, a failed pool.
3. Cold archive - data you’re keeping because deleting it feels wrong, not because you access it. Finished projects, old media rips you’re unlikely to rewatch soon, footage, anything you’d be sad to lose but wouldn’t notice missing for months. This is where tape belongs.

Trying to make tape your only backup is a mistake, because restoring from tape is slow and sequential (finding one file means shuttling the tape to the right position, not an instant seek) and because a single tape drive is itself a single point of failure with no redundancy unless you buy a second one. Tape earns its keep as the layer that takes cold data off your spinning disks entirely, freeing up array space and cutting the power bill for drives that were only spinning to hold data nobody was reading anyway.

LTFS vs real backup software

You have two broad approaches to actually writing to tape:

LTFS (Linear Tape File System) presents the tape as something closer to a regular filesystem, letting you drag-and-drop files onto it with normal OS tools once it’s mounted. It’s the easiest on-ramp for a home user and works fine for straightforward “copy this folder to tape” archival. The tradeoff is that LTFS doesn’t give you incremental backups, verification, or a searchable catalog of what’s on which tape. You’re the catalog.

Dedicated backup software (Bacula and Amanda are the two open-source options homelabbers actually reach for; Veeam and Commvault dominate on the commercial/enterprise side but are overkill and expensive for a home setup) treats tape as a proper backup target: it tracks what’s on which tape in a database, handles incremental jobs, verifies writes, and can span a backup job across multiple tapes automatically when one fills up. This is meaningfully more setup work than LTFS, but it’s the difference between “I have a pile of tapes and a spreadsheet” and “I can actually find and restore a specific file from six months ago without guessing which tape it’s on.”

For anything past a handful of tapes, use real backup software. The point where LTFS’s simplicity stops paying off and its lack of cataloging starts costing you time is lower than most people expect, usually somewhere around tape number five or six, once you can no longer just remember what’s where.

The gotchas that catch people

Used drives are a real gamble. The used-enterprise-gear market (see the buying used enterprise gear guide for the general version of this advice) applies doubly to tape drives, because a tape drive has more precision moving parts than almost anything else in a homelab, and a worn read/write head or dirty tape path can silently produce bad writes that look fine until you try to restore. Buy from a seller with return policy, and run a full write-then-verify pass on the first few tapes before trusting the drive with anything you can’t lose.

Generational compatibility is narrower than people assume. An LTO drive can typically read two generations back and write one generation back (an LTO-8 drive reads LTO-6 and LTO-7, writes LTO-7 and LTO-8, but can’t touch LTO-9 media at all). Buying an odd mix of drive and tape generations off eBay without checking this table first is a common way to end up with tapes a drive can’t use.

It’s genuinely loud and it runs hot under load. These are enterprise mechanisms built for datacenter racks, not quiet home offices. Plan for where it physically lives accordingly, the same consideration that applies to used enterprise gear generally.

Tape is not immune to failure, it’s just failure on a different, generally more forgiving timeline. Tape media has a real, if long, shelf life (LTO media is commonly rated for 15-30 years under proper storage conditions: cool, dry, away from strong magnetic fields), and a specific tape can still fail to read. It’s not a substitute for having more than one copy of anything irreplaceable, it just makes the “more than one copy” math a lot cheaper at scale.

Air-gapped, and that’s a real security feature, not a nostalgia bonus. A tape sitting on a shelf, unmounted, is completely unreachable by ransomware, a compromised backup server, or a bad automation script that decides to delete things. That’s a genuinely different threat model than any always-connected backup destination, disk, cloud, or otherwise, and it’s part of why enterprises never fully abandoned tape even with cloud cold storage tiers available.

Is it actually worth it for you

Be honest about your actual data growth and access patterns before buying anything. Tape rewards people who have accumulated a real, non-trivial amount of genuinely cold data (measured in tens of terabytes, not a handful) and who don’t mind a slower, more deliberate archival workflow in exchange for lower long-run cost and a truly offline copy. It punishes people who buy a drive expecting it to behave like a slow external disk, or who don’t have enough data yet to amortize the drive cost.

If that’s not you yet, keep building out disk-based redundancy and off-site backup first. Tape is the tier you add once those are solid and your cold-data pile has grown past the point where another external drive is the obvious answer.