A well-stored magnetic tape preserves its data for decades: it is almost never the medium that dies. The real risk is no longer being able to read it. To extract the contents of an old LTO, DLT or DAT tape, you need to bring together a working drive of the right generation, the right connectivity (SCSI or SAS), a compatible computer and operating system, and the software able to understand how the backup was written. It is this entire ecosystem that disappears year after year. The right benchmark: beyond fifteen years, it becomes urgent to extract the data you will still want to use tomorrow.

This article explains why a perfectly readable archive can become inaccessible, and how to migrate before it is too late.

The tape, a surprisingly durable medium

Let us start with the good news. A magnetic tape (LTO, DLT, DAT) stored in good conditions withstands time remarkably well.

These good conditions are well known: sheltered from light and ultraviolet, in a room with a cool and stable temperature of around 18 to 20 °C, without significant thermal swings, and protected from humidity. A tape kept this way keeps its data readable for decades. Physical degradation of the medium is almost never what brings us a case.

What causes problems, then, is not the longevity of the tape, but everything that has to be assembled around it to extract the contents. This plays out on two fronts: hardware and software.

First obstacle: finding a compatible and functional drive

A tape does not read itself. You need a drive, and not just any drive.

Generation compatibility. Each tape generation requires a suitable drive. The LTO consortium's rule is precise: a drive reads the tapes of its own generation and of the two previous generations, but writes only to its own generation and the previous one. An LTO-5 drive, for example, reads LTO-3, LTO-4 and LTO-5 tapes. Beware, however: this rule was tightened from LTO-8 onward, which reads only the immediately preceding generation. The older your tapes are, the narrower the window of drives able to read them, and these drives are becoming scarce on the market.

A drive that really works. Finding the right model is not enough. Some drives require a cleaning cartridge and stubbornly refuse to read a single tape until they consider themselves clean. An "available" drive is not necessarily an operational one.

Connectivity, a headache that keeps getting worse. This is often where everything gets complicated. If you are lucky, the drive uses a SAS interface (Serial Attached SCSI), a recent technology that connects without difficulty to a modern computer. But many old drives use parallel SCSI, and there is not one SCSI but a whole succession of standards: SCSI-1, SCSI-2, Fast and Wide SCSI, Ultra SCSI, Ultra2 in LVD, up to Ultra160 and Ultra320. Each has its connectors (Centronics 50, HD50, HD68, VHDCI) and its termination rules (passive, active, LVD).

The older the standard, the more you need a period controller card, and a motherboard able to accommodate that card. Yet old SCSI cards are becoming rare, and recent motherboards do not recognize them. Reconstituting a complete chain sometimes resembles an archaeological dig: a compatible drive, connected to the right controller card, itself detected by a sufficiently old motherboard, on an operating system where the right drivers could be installed. Only at the end of this chain does the tape become readable.

Second obstacle: the format, the software and the catalog

Let us assume the hardware problem is solved: the drive is running, the computer sees it, the tape is spooling. A new wall rises up, this time on the software side.

You have to understand how a tape is written. Originally, it is completely blank, with no structure at all. The drive begins by inscribing its own organization, specific to the drive model, by placing markers called file marks, which delimit the start of the tape, the start of a session, and so on. On top of this structure, the backup software finally lays down the data.

And everything depends on that software. In the best case, it used an open and documented format that we know how to interpret: MTF (Microsoft Tape Format), TAR, or IBM formats. In the worst case, the format is proprietary, sometimes encrypted: you then need the original software, not only to open the tape, but to understand how the backup was built on it.

The central role of the catalog. A backup catalog is the index maintained by the software: it records which files were backed up, when, on which tape, and in which successive versions. Without this catalog, you do not know what the tape contains. You then have to read it in full, copy the contents, and reconstruct the index.

Hence two routes for extracting your data:

Extraction route	When	How
Original software	The period backup software is available	We recreate the catalog with the original software, then extract the files (we keep a large stock of legacy software)
Direct extraction	Software unavailable or format uncooperative	We use our own tools or a custom-developed analysis to extract the data on the fly, into a readable directory tree

You still have to know which software you are dealing with. Very often, the client no longer has any idea what was used to write their tapes. We then analyze the start of the tape in hexadecimal, looking for recognizable patterns that reveal the software used at the time. This identification determines everything that follows.

A true story: when recovering a tape means digging out a 20-year-old Mac

One case illustrates the difficulty well. A batch of tapes reaches us; on analyzing the start of the contents, we identify the software: Retrospect.

The problem is like Russian nesting dolls. Retrospect's format is not open: its structure is opaque, and the data is compressed according to methods that cannot be guessed from the outside. To read this backup, you therefore need the Retrospect software itself. But Retrospect only runs on Mac. And each version of Retrospect produces different backups: you therefore need exactly the right version of the software. Now that particular version requires a specific version of macOS, which itself requires a specific model of Apple computer.

Step by step, recovering this data required digging out old Apple computers about twenty years old. With a fair amount of effort, the client's data was ultimately extracted. The anecdote sums up the whole point of this article: a perfectly readable tape can remain inaccessible for years, simply because the software chain that goes with it has disappeared.

When should you migrate? The 15-year benchmark

There is no cut-off date, but there is a logic. Migration must be a considered process: identifying, among your tapes, those whose contents must remain usable in the future, and dealing with them first.

The guiding principle is not to wait too long. Every year that passes complicates the task, and not only because of the hardware. Internal knowledge is lost: the IT manager of the day retires, the drives are scrapped, the old computers thrown out. The moment comes when no one in the company knows any longer what the tapes contain or how to read them, and we are handed a box of cartridges with, as the only indication: "we are not really sure what is on it."

In practice, fifteen years is a good benchmark for extracting the contents of tapes whose future use you want to preserve.

Which data deserves this attention? First, the most sensitive and most regulated: medical records, banking data. Then anything that retains value over time, such as patents. But also a heritage that is often forgotten: weather records spanning thirty or forty years, photographic archives. Such data is irreplaceable: losing the tape means losing forever decades of measurements or visual memory.

The regulatory dimension is not minor either. Some data is subject to long legal retention periods, particularly in banking and healthcare. A medical record, for example, must be kept for many years, and remains useful well beyond: in the event of death, it makes it possible to trace the operations undergone, the implants or devices fitted, the medical history. So much information that has value only if it can still be read.

How an extraction unfolds, step by step

Concretely, here is how we proceed with a batch of old tapes.

1. Reconstituting the hardware setup. We assemble the appropriate reading chain: the drive compatible with the tape's generation, the right controller card (SCSI or SAS), a period operating system fitted with the necessary drivers.
2. Bit-by-bit copy. With our in-house software, we make a raw copy of the tape. This step already determines the block size (fixed or variable) and allows the first 100 to 200 MB to be dumped.
3. Content analysis. On this extract, we look for known patterns to identify the format: MTF, IBM formats, TAR, or other.
4. Quote and mandate. The analysis leads to a quote and a mandate that set the rest of the work.
5. Extraction. Depending on the identified format, we work either directly from our copy to extract the data on the fly, or, in specific cases, with the original backup software. The result is a structured and readable restitution of your data.

Frequently asked questions

How long does a magnetic tape last?

Well stored (sheltered from light, at a stable 18-20 °C, without humidity), an LTO, DLT or DAT tape preserves its data for decades. The practical limit is not the lifespan of the medium, but the availability of a drive, connectivity and software still able to read it.

Why can an old backup tape no longer be read?

Because an entire chain has to be reconstituted: a working drive of the right generation, the appropriate controller card (often old SCSI), a compatible motherboard and operating system, and the original software or its catalog. A single missing link is enough to make the tape inaccessible.

Can a recent LTO drive read my old LTO tapes?

According to the LTO consortium's rule, a drive reads its own generation and the two previous ones, and writes to its own generation and the previous one. An LTO-5 therefore reads from LTO-3 to LTO-5. This compatibility was reduced from LTO-8 onward, which reads only the immediately preceding generation.

What is a backup catalog and why is it essential?

It is the index created by the backup software: it indicates which files were recorded, when, on which tape and in which versions. Without it, you do not know what the tape contains; you then have to read it in full and reconstruct this index before any extraction.

After how long should you migrate your tapes?

A good benchmark is around fifteen years, dealing first with regulated data (medical, banking), patents and irreplaceable archives. The decisive factor is not to wait until the drives, the software and the internal skills have been lost.

Can a tape whose software is no longer known be recovered?

Yes. We analyze the start of the tape in hexadecimal to identify the format and the original software from recognizable patterns, then extract the data with the period software or with custom-developed tools. Even a box of "unknown" tapes can be analyzed.