Tape storage is, in some sense, the most successful technology across all of IT, being first used in the 1950s and AFAIK still unbeaten today in its niche. Though the success is mostly because tape can be wound up so compactly, it also seems to be intrisically quite good to handle – unlike with big drums or disks, you don't need to carefully align the heads on a huge surface but can just let a small part of the tape come to the fixed head. And at any rate, the tape head technology had been well optimised since the beginning of audio recording.

Yeah, sure tape is also famous for its rip- and knottability, needs careful alignment too etc., but I think we can say it's an overall very good medium.

Spooled-up tape has the obvious problem of access time, which has always pretty much confined it to backups only, however that's not the only way to operate a tape – it's also possible to close the tape to a loop, so each part periodically passes the head. This isn't only theoretical, it has been the standard way to achieve echo in audio production. And we might actually say that a drum storage is nothing but a bunch of closed rigid tape loops in parallel, so if drums worked well for storage then tape loops should have worked well to. Except they would presumably have been cheaper, and the loops could be made much longer than a drum circumference, to trade in latency for capacity – at arbitrary ratio!

This would certainly not have been as cheap as spooled tape because to get really high capacity you'd still need multiple loops, each with its own head. But that too is actually something that was done in audio, in the Mellotron. At least after audio cassettes became a consumer mainstay, the heads should have been cheap enough to make this viable.

What am I missing?

  • That's an interesting idea. It's similar in concept to mercury delay lines, which there is a question about on here already, but with much greater latency and capacity. – Wilson Aug 17 at 16:54
  • 6
    This is what Sinclair thought in designing the Microdrive — en.m.wikipedia.org/wiki/ZX_Microdrive — but alas build quality for that particular instance was, ummm, questionable. – Tommy Aug 17 at 16:59
  • 6
    There was the stringy floppy too... – Stephen Kitt Aug 17 at 17:24
  • 4
    @StephenKitt I was about to answer with the Exatron Stringy Floppy, until I saw that you already posted it as a comment. Turn it into a real answer. – manassehkatz Aug 17 at 18:50
  • 1
    Worth noting that continuous-loop tape was a popular data storage medium for several decades in at least one application (aircraft flight recorders). – Sean Aug 18 at 21:53
up vote 27 down vote accepted

Because a closed tape loop doesn't buy you anything in random access latency, which is presumably the problem you're trying to solve.

The problem is that large tape loops, like 8-track, only can wind one way. So on average, you need to wind the tape halfway through to get to a random piece of data. The worst case, you wind the entire tape.

Random access tape (IBM 3570 and the canceled LTO Accelis) keeps the tape at the midpoint by default. So in the average case, you wind a quarter of the tape, in the worst case you wind half the tape. This is a 2x improvement.

If you have a bi-directional tape loop, you can avoid this, but then you have a density problem because the tape is no longer wrapped in a spool, but on a series of rollers.

The fundamental problem with tape is that in order to get decent transfer rates, you need to move the entire reel of tape very fast, which is difficult, or spin the heads independently of the tape (helical scan), which brings its own set of alignment issues. A 7200 RPM hard drive gives you >20 m/s linear velocity, exceeding tape (~7 m/s for LTO).

Once you have the tape moving, the heads are contacting the medium, so it will wear out with every pass. Back when we used helical-scan DDS tape, it was rated for 2,000 passes, requiring 10 passes per read, so the tape was only good for 100 full read or write cycles.

Hard drives heads float on an air bearing, making their lifetime much greater and giving a much more stable head-disk interface. Because there's very little friction in the system, you can very precisely servo the heads to a track. Tape doesn't let you do this. The best tapes today (LTO-6) give you about 13,000 tracks per inch, where as hard drives exceed 400,000 tracks per inch, a 30x increase.

  • 11
    I reckon “it will wear out with every pass” is the crucial point – it explains why short tape loops that keep looping around every few seconds (which is what I had mainly in mind with the question) don't work out, at least not in a 24/7 application. – leftaroundabout Aug 17 at 22:11
  • @leftaroundabout: a short tape loop doesn't have much storage per motor + heads. Seems unlikely to be of any use compared to a hard drive with ~200x better random access latency (~10ms vs. ~2000ms for a 4 second tape loop). As this answer points out, the rigidity and other factor of an HD allow much higher areal density for a similar era of technology. Were you imagining that even a short tape loop would have a useful amount of storage? If it repeats every 4 sec, and can read/write at 200MB/s, you have at most 800MB of storage (unless you very wide multi-track tape?) – Peter Cordes Aug 18 at 15:46
  • though @leftaroundabout there were tape formulations that never had oxide contact with the heads and drive gear. DECTape was mylar-coated on both sides and incredibly durable — if not very high density. – scruss Aug 18 at 17:03
  • 1
    @PeterCordes there was a time when a 800 MB storage with “only” 2 s latency would have been amazingly good. Hard drives only started their journey towards today's storage density and reliability when the industry figured out flying heads and giant magnetoresistance. – leftaroundabout Aug 18 at 23:08
  • @leftaroundabout: and back when 800MB total storage would have been very good, 200MB/s read/write bandwidth was unheard of for hard drives, and I assume for tape drives. Because of the way your proposed design works, your storage capacity always = latency x bandwidth product (times 2), unless like I said you had multiple tracks on one tape you could seek in 2 dimensions. 200MB/s is typical for 7200RPM magnetic HDs made in the last couple years. Older HDs were much slower. – Peter Cordes Aug 19 at 4:15

They were used, but suffered from latency issues and complex electro-mechanical design challenges. Their brief day in the sun was around the mid-1950s: after that, it was likely cheaper to add another drum or disk platter rather than work out massive tape-loop logistics.

An example of a large tape-loop system was the Machine Mathématique IRSIA-FNRS (MMIF), built in Belgium in 1955-56:

The auxiliary memory consisted of six tape loops, each 100 meters long, and of a capacity of about one Mbits (10 bits per mm). The tapes stored data from input coding devices and for output printing machines. Each tape was wound around a hollow driving barrel which was covered by a cap. The tape movement was controlled by a pneumatic clutch. Air valves controlled the relative air pressures in the barrel and in the cap. A depression in the barrel pressed the tape against the barrel and drove it; on the contrary, an overpressure in the barrel – i.e. a depression in the cap - applied the tape against the cap and stopped it. Coming from a driving speed of about 1m/sec to a brutal halt and vice-versa took no more than about 2,5 msec

(quoting from Courtois PJ. (2011) The Belgian Electronic Mathematical Machine (1951-1962): An Account. In: Jones C.B., Lloyd J.L. (eds) Dependable and Historic Computing. Lecture Notes in Computer Science, vol 6875. Springer, Berlin, Heidelberg)

Later home-computer tape loop systems, such as the Exatron Stringy Floppy and the Sinclair Microdrive, only came into existence after floppy drive technology had been developed. They were only viable while they offered adequate performance for a far lower price than add-on floppy drives. Once floppies were cheaper, tape loop drives were quickly forgotten.

The University of Strathclyde had a large fleet of Sinclair QLs that it loaned to its students. These were perhaps the only machine that initially relied entirely on Microdrive tape loops. The pragmatic approach to unreliable tape media was to encourage students to buy more tapes, format them, and return the unreliable ones for refund. Since Microdrive tapes stretched as they were used, their capacity increased up to a generally stable value. Accepted practice at the time was to format tapes up to 10 times using a FOR loop in SuperBASIC. If a tape stayed at the same capacity over the first few formats, it was likely to be reliable. If, however, the capacity kept increasing it meant that the tape was likely to break and that particular cartridge should not be used.

  • Beyond the Exatron Stringyfloppy and the Sinclair Microdrives was also the Rotronics Wafadrive that tried to compete with the Sinclar Microdrives at a slightly lower price for a dual drive. They offered "wafas" between 16 and 128k of capacity. It wasn't very successful, being slower and less reliable than Sinclair's original. – tofro Aug 17 at 23:10
  • 2
    Less reliable? Whoa, that would take some doing! – scruss Aug 18 at 0:46
  • The drives themselves were built like a tank - The tape cartridges were the problem, binding and jamming and eventually simply ripping the tape. – tofro Aug 20 at 7:05
  • @tofro - "Beyond the Exatron Stringyfloppy was also the Rotronics Wafadrive". Though "the Wafadrive comprised two continuous loop 'stringy floppy' tape drives" (Wikipedia), much as "Texas Instruments licensed the Stringy Floppy as the Waferdrive" (Wikipedia). – Nigel Touch Aug 20 at 14:42

unlike with big drums or disks, you don't need to carefully align the heads on a huge surface but can just let a small part of the tape come to the fixed head.

Sounds a bit like you never had to do so ;) I started my professional life in tape/disk service And believe me, there's a lot to be adjusted at a 34xx compatible tape drive head - not to mention what's needed to adjust the capstan drive and the vacuum chambers :)) (*1)

Spooled-up tape has the obvious problem of access time, which has always pretty much confined it to backups only,

Not really. that's maybe basic use of today – and what modern tape systems are optimized for. But there was a time when tape drives were a major component of external storage. In many cases the only one. If at all, disks where used for OS and software and maybe paging. All data was done with tapes (*2).

The heydays of tapes as main data media lasted through the whole 1970s with the IBM 3420 as cornerstone.

A 3420-8 (the top model in 1973) could read, write and seek data at 200 inch per second at 6250 bpi - which, at 9 track was also a byte. This gave a raw speed of up to 1,25 MiB/second. A maximum sized spool of 2400ft could hold 100..130 MiB (depending on block size) and be read/written/seeked in about two minutes. Pure rewind was even shorter at ~45 seconds. Further, blocks could be read in either direction.

These tapes were not meant as backup, but work media. Usually not the tape, but the mainframes CPU and application speed was limiting throughput and when used for disk backup, the tape had to wait for disk seek.

These tapes were blockwise rewritable (when keeping the same block size). From an applications view the differences to disk drives where rather minor - when keeping the tape nature in mind.

however that's not the only way to operate a tape – it's also possible to close the tape to a loop, so each part periodically passes the head. This isn't only theoretical, it has been the standard way to achieve echo in audio production.

And as well for cheap microcomputer storage with the stringy floppy and its most well known incarnation, the ZX Microdrive. A bit like the floppy story, as mainframe drives were far ahead.

Except they would presumably have been cheaper, and the loops could be made much longer than a drum circumference, to trade in latency for capacity – at arbitrary ratio!

What am I missing?

Beside that the fact that it has been done, that there is no speed gain.

If the tape is longer than a few meters, it needs to be spooled which in all cases will make it only single direction, resulting in exactly the same worst case access times as a bidirectional read standard tape, while being less effective in average seek time. Unless speed is greatly reduced, the mechanics would have needed the same effort as 3420 drives spend, thus being as expensive, without any improvement. With more simple mechanics, speed had to be greatly reduced to keep wear in check. Existing (cheap) stringy floppies did offer only 1/100 of the read sped of a 3420-8

Bottom line: What you wanted to do with loops was already achieved with standard tapes. They were a great storage (when used right), but did need exceptional effort to do so - and modern backup ones aren't in line with these requirements.


*1 - The Wiki got a rather good explanation of the workings in their 9-Track-Tape article - even though it's tied to a medium speed drive with vacuum chambers, but a conventional capstan.

*2 - I remember several customers who even in 1980 refused to switch for disk storage. The mainframe could be booted from tape and as data storage tapes could work at similar speeds as disks. Given a useful data organization ofc.

  • Interesting to hear from experience. No, I have indeed not adjusted any storage tape myself – all tapes I've used were various audio machines. – You haven't really addressed the main idea behind my question: really short tape loops, only like a meter, unspooled, a whole lot of them in parallel. I accepted user71659's answer which suggests the problem would be that tape wears out with each pass at the head, can you confirm this? – leftaroundabout Aug 17 at 23:08
  • @leftaroundabout An answer like you want to get is way out of scope as it would require a description of a 3420 tape drive in detail. Waht you know are acoustic taped pressed with force onto a head and moved by friction. A 3420 uses a sysem of vacuum chanbers to level out tape movement for short distance transport, while also holding the tape onto the head with almost no pressure against the head, moved by capstans that hold the tape not by friction but again byvacuuom, again puting no stress on the tape.This way a tape could passed thens of thousends of times across a head without much wear. – Raffzahn Aug 17 at 23:21
  • @leftaroundabout To make the same working with short loops woudl mean inccoporating the same effort as for a with a 3420 tape drive - and having many is like having many tape drives. So why just storing ~250 KiB per meter loop and drive when it's possible to have many megabytes with the same effort? Real tapes in the 1970s where already way faster and better than what could have been archieved with your solution - and even at a loweer price than your loops. Keep in mind, back then and with a right (real world) setup, tapes easy outperformed disks. – Raffzahn Aug 17 at 23:23
  • Or in simple words, it would have been an invention without use case and market, as it would have brought no performance gain or money savings. – Raffzahn Aug 17 at 23:31
  • 1
    Since tapes at that time were blockwise rewritable and /dev/mt0 on UNIX systems was a block device, you could put a filesystem on a tape and mount the filesystem. And write to it. If you were patient. A lot of operations result in writing to an inode, so the drive was constantly seeking to the beginning of the tape. – Mark Plotnick Aug 18 at 0:12

Closed tape loops may not have been popular, but they were used occasionally.

Colossus tape unit

One of the first computers, Colossus, used a paper tape loop for read-only data storage (the ciphertext to be decoded was stored on the tape). This tape was stuck in a loop and driven at high speed (~45 km/h) past an optical reader.

There are two ways to achieve a tape loop:

  1. Keep all of the tape perpendicular to the same plane, and add enough rollers that the tape never has to wrap against itself. This approach works well for things like tape echo units that have only a few seconds of tape, if even that much, but quickly gets bulky and unworkable for handling larger quantities. If tape were stored in a cartridge that's 4cm on a side, it would be necessary to have about ten rollers for each meter of tape. Perhaps not a problem if the loop is short, but a big problem if it isn't. An advantage of this approach is that the tape can be kept under tension.

  2. Have the tape loosely coiled up in the cartridge around a spool, and pull tape upward from the middle. This approach can work (it's how 8-track tapes are constructed), but it requires that the tape slip against itself in the cartridge (causing excessive wear) and is prone to have problems if things don't wind and unwind themselves just right.

To achieve good reliability and capacity in a tape cartridge, tape needs to be coiled and uncoiled while staying in a common plane. Non-looping cartridges can do that. Looping cartridges may allow for slightly-simpler recording/playback equipment, but their downsides are too great to make them particularly competitive with other technologies.

  • ZX Microdrives also take approach 2 above. – Toby Speight Aug 20 at 7:59

An erstwhile popular closed tape loop analog storage medium (8-track tape) was used to carry digital signals, although only for toy purposes:

Milton Bradley's (MB) OMNI Entertainment System was an electronic quiz machine game first released in 1980, similar to Jeopardy! or later You Don't Know Jack video game series, using 8-track tapes for playback analog audio for questions, instructions and answers as well as digital signals in magnetic tape data storage [emph. mine - LeoB.] on remaining tracks to load the right answer for counting the score. In 1978, the Mego Corporation launched the 2-XL toy robot, which was similar.

So, formally, there was a popular storage medium made of a closed tape loop which was used as data storage, albeit not for computing-related applications, and read-only.

https://en.wikipedia.org/wiki/ZX_Microdrive

"Microdrives used tiny (44 × 34 × 8 mm including protective cover) cartridges containing a 5-metre (200-inch) endless loop of magnetic tape, 1.9 mm wide, driven at 76 cm/second (30 in/second); thus performing a complete circuit in approximately 8 seconds."

https://en.wikipedia.org/wiki/Exatron_Stringy_Floppy

https://en.wikipedia.org/wiki/Rotronics_Wafadrive

The former ZX owners would tell you stories about how unreliable these tapes were. Compared to audio compact cassette players (TRS-80 !), at least you didn't have to press the play-stop-forward-reverse-record buttons.

Your Answer

 

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.