The world of tape backup was (is?) absolutely filled with all sorts of vendor-lock in projects and tools. It's a complete mess.

And even various versions of tar aren't compatible, and that's not even starting with star and friends.

It's not just limited to tape, most archiving and backup software is proprietary. It's impossible to open Acronis or Macrium Reflect images without their Windows software. In Acronis's case they even make it impossible to use offline or on a server OS without paying for a license. NTBackup is awfully slow and doesn't work past Vista, and it's not even part of XP POSReady for whatever reason, so I had to rip the exe from a XP ISO and unpack it (NTBACKUP._EX... I forgot microsoft's term for that) because the Vista version available on Microsoft's site specifically checks for longhorn or vista.

Then there's slightly more obscure formats that didn't take off in the western world, and the physical mediums too. Not many people had the pleasure of having to extract hundreds of "GCA" files off of MO disks using obscure Japanese freeware from 2002. The English version of the software even has a bunch of flags on virustotal that the standard one doesn't. And there's obscure LZH compression algorithms that no tool available now can handle.

I've found myself setting up one-time Windows 2000/XP VMs just to access backups made after 2000.

I can only speak for macrium but they have good reasons to use their own format, so that you can have differential mountable backups. That's very different from someone inventing tar-but-worse.

I have at various times considered a tape backup solution for my home, but always give up when it seems every tape vendor is only interested in business clients. It was a race to stay ahead of hard drives and oftentimes they seemed to be losing. The price points were clearly aimed at business customers, especially on the larger capacity tapes. In the end I do backup to hard drives instead because it's much cheaper and faster.

Tape absolutely isn't viable for the consumer at all, but definitely worth exploring for the novelty. Even if you manage to get a pretty good deal on a legacy LTO system (other formats don't even come close to the tb/$ of 10+ year old LTO and drives are still fairly cheap), the drives aren't being made any more and aren't getting any cheaper. Backwards compatibilty may be in your favor depending on your choice of tape generation at least, I think there's at least two generations guaranteed. Optical will probably remain king though the pricing is worse than HDDs, there's no shortage of DVD or BD readers, but you might run into issues with quad layer 128 BD as they only hit the market fairly recently.

The only reasonable solution is to keep migrating and checking the data on various media; but this is expensive and often deemed not worth it.

Depends on the size of the data.

If your dataset is below 1-2Tb it would cost you less than $200 in a decade to move to a newer HDD every 5 years.

That used to be my method but I decided to "upgrade" to paying backblaze to manage that method for me and just in the nick of time too as my SSD crib-deathed soon after.

Tape drive and Bareos/Bacula "just works"

Absolutely not worth it tho. Drives are hideously expensive which means they only start making sense where you have at least dozens of tapes.

There is an advantage of tapes not being electrically connected most of the time so lightning strike will not burn your archives, I have pondered making a separate box with a bunch of hard drives that boots once a month and just copies last months of backups on hard drives, powered from solar or something just to separate from the network

The only way to do tape at home is with used equipment and Linux/BSD. You can do quite a bit with tar and mt (iirc) - even controlling auto loaders.

What’s fun are the hard drive based systems designed to perfectly imitate a tape autoloader so you don’t have to buy new backup software (virtual tape libraries).

ARCServe was a Computer Associates product. That's all you need to know.

It had a great reputation on Novell Netware but the Windows product was a mess. I never had a piece of backup management software cause blue screens (e.g. kernel panics) before an unfortunate Customer introduced me to ARCServe on Windows.

My favorite ArcServe bug which they released a patch for (and which didn’t actually fix the issue, as I recall) had a KB article called something along the lines of “The Open Database Backup Agent for Lotus Notes Cannot Backup Open Databases”.

IIRC tar has some Unixisms that don't necessarily work for Windows/NTFS. Not saying reinventing tar is appropriate but there's Windows/NTFS that a Windows based tape backup need to support.

Most of what makes NTFS different than FAT probably doesn't need to be backed up. Complex ACLs, alternative data streams, shadow copies, etc, are largely irrelevant when it comes to making a backup. Just a simple warning "The data being backed up includes alternative data streams. These aren't supported and won't be included in the backup" would suffice.

All of that stuff matters when you're using the backup for its intended purpose: to restore a system after hardware failure.

Unix tar is obviously not the right solution, but a Windows tar seems like it shouldn't be that hard to do and yet we are in the situation we are today. I've been using dump/restore for decades now on Unix, including to actually recover from loss, but I admit that it's not that pleasant to use. I like that it is very simple and reliable however, unlike the mess that is Time Machine (recovering from a hardware loss on a Mac is a roll of the dice, and I've gotten snakes) or worse Deja Dup. I'm not sure I've ever successfully recovered a system from a Deja Dup backup.

> using the backup for its intended purpose: to restore a system after hardware failure.

No. The intended purpose of a backup is to restore the data (such as the Frogger 2 source code) after a hardware failure. If it has the side effect of also producing a working system, that's good, but it's not the point. After all, the hardware necessary to build a working system may not exist any more; one (only-probably not the last) instance of said hardware just broke, after all.

Your one trivial use case isn't all use cases, and it sure isn't my important one. If you're doing more than backing up your personal workstation, metadata is extremely important. If I ever have to restore something, even just data, out of the multi-petabytes we have on tape, I better not have to manually go through it to figure out who should actually have access to it before I make it available to the people who need it.

The metadata matters.

I think the use case for disaster recovery is a bit different than long-term archival.

Does anyone buying tape storage actually use it for archival rather than disaster recovery?

We do. We have vast amounts of data that we merely have to keep around for a decade or so for compliance purposes that would rarely be accessed, so it goes on tape and off to Iron Mountain. Then we have backups where we need to be able to recover a running system from some 'known good' state, which is somewhat complicated. The former is conceptually tape drive+tar/cpio/etc.; the latter is an expensive setup that includes some proprietary solutions.

If you’re backing up a db or something sure, but for a file server this can be just as important as the data itself (ex: now everyone can read HR’s personnel files which had strict permissions before)

That’s fair; I wasn’t really considering windows. It seems like there ought to be some equivalent by now though.

The format is extensible enough that it could be added

The company that made it probably was hoping for vendor lock-in

Vendor lock in for backup and archival products is so ridiculous. It increases R&D to ensure the lock-in, and the company won't exist by the time the lock-in takes effect.

Well yes, but the boss probably is willing to invest more money (meaning higher salaries, more people, better tools) expecting a future return than when using reasonable formats.