4 ms·

I think our best bet would be tape. Currently tape is amazing technology for archiving data and long term storage. The pysical media costs very little...the bi

I think our best bet would be tape. Currently tape is amazing technology for archiving data and long term storage. The pysical media costs very little...the big cost is in the hardware for reading and writing. I really wish we had an alternative open hardware competitor to the current tape market. I could justify the cost of tape but even used a lto7 library or a single reader for a desktop can run from a few grand for parts all the way up to 8 grand from what iv seen. YMMV

vlovich123 3y ago
Is there something about tape HW itself that’s inherently complex and expensive? Or is it just that the target market is enterprise and so the manufacturers are able to price discriminate more effectively? Regardless, I doubt tape is a good consumer solution because it’s primarily for archival. That means that retrieving data is super slow (average seek time is ~1 minute). On the other hand, given that’s for an 18TiB cartridge, you might be able to cut that drastically by using less capacity. So it would be interesting if someone attempted a consumer play with tape, I just don’t know what the business model would look like since most customers rely on the cloud for archival to offload the technical and physical maintenance parts of the problem.
- toast0 3y ago
  It's probably a mix. Consumer tape drives used to exist, but uptake was low. And really, most consumers want to ignore backups and hope nothing bad happens (which usually works out ok). Something like a cost reduced two generations old LTO tape drive might work, but of course it has to compete with surplus two generations old LTO tape drives and be careful not to attract the business of enterprise which should be buying the current drives. Anyway, I'm not sure how much cost reduction is possible --- there's a lot of precision movement required. Edit: also, at least in the current market, I doubt many consumers have enough data to make tape economical. An online array of rotational drives, plus backup to and rotation of external drives is probably a better fit for consumers, and a lot of enterprises.
sschueller 3y ago
Tape would be perfect because people don't tend to skip through movies to random points. At least I don't.
- toast0 3y ago
  You might not, but some people do. And disc based formats assume reasonable seek times for menus. Plus if you're storing multiple movies on a tape, you still need to seek to the beginning of the one you want. A 100-disc changer or an archive of discs onto rotational hard drives would be a better experience.
  - jasomill 3y ago
    Seek time wouldn't necessarily be a deal killer, as you could store chunks of movies as files in a structure similar (or identical) to LTFS[1] and have average seek times from tape load less than a minute and local seeks (forward/back chapter/minute/etc.) somewhere between a few seconds and instantaneous, depending on chapter length and how much cache RAM you're willing to throw at the problem. Bear in mind that tape movement in an LTO drive is many times faster than a naïve tape capacity / transfer rate calculation might suggest, as LTO drives require anywhere from a few dozen to a couple hundred head passes to read/write a full tape, and seek to a given logical block number considerably faster than they can read or write. The bigger problem I see with tape library vs. RAID for consumer movie storage is cost. Even if we assume economies of scale would drive the cost of LTO-9 drives down to, say, $1,000, you'd still need to store a lot of movies for the cost of LTO-9 drive + library robotics + tapes to approach the cost of large capacity hard drives + additional cabling and interface electronics, given that $1,000 will buy enough hard drive capacity to (non-redundantly) store over 1,000 50 GB UHD Blu-ray movies. [1] https://en.wikipedia.org/wiki/Linear_Tape_File_System https://en.wikipedia.org/wiki/Linear_Tape_File_System
    - sschueller 3y ago
      Is the cost of LTO due to the technology or the small market and a few large players taking advantage of that?

vlovich123 3y ago

Is there something about tape HW itself that’s inherently complex and expensive? Or is it just that the target market is enterprise and so the manufacturers are able to price discriminate more effectively? Regardless, I doubt tape is a good consumer solution because it’s primarily for archival. That means that retrieving data is super slow (average seek time is ~1 minute). On the other hand, given that’s for an 18TiB cartridge, you might be able to cut that drastically by using less capacity. So it would be interesting if someone attempted a consumer play with tape, I just don’t know what the business model would look like since most customers rely on the cloud for archival to offload the technical and physical maintenance parts of the problem.

toast0 3y ago
It's probably a mix. Consumer tape drives used to exist, but uptake was low. And really, most consumers want to ignore backups and hope nothing bad happens (which usually works out ok). Something like a cost reduced two generations old LTO tape drive might work, but of course it has to compete with surplus two generations old LTO tape drives and be careful not to attract the business of enterprise which should be buying the current drives. Anyway, I'm not sure how much cost reduction is possible --- there's a lot of precision movement required. Edit: also, at least in the current market, I doubt many consumers have enough data to make tape economical. An online array of rotational drives, plus backup to and rotation of external drives is probably a better fit for consumers, and a lot of enterprises.

sschueller 3y ago

Tape would be perfect because people don't tend to skip through movies to random points. At least I don't.

toast0 3y ago
You might not, but some people do. And disc based formats assume reasonable seek times for menus. Plus if you're storing multiple movies on a tape, you still need to seek to the beginning of the one you want. A 100-disc changer or an archive of discs onto rotational hard drives would be a better experience.
- jasomill 3y ago
  Seek time wouldn't necessarily be a deal killer, as you could store chunks of movies as files in a structure similar (or identical) to LTFS[1] and have average seek times from tape load less than a minute and local seeks (forward/back chapter/minute/etc.) somewhere between a few seconds and instantaneous, depending on chapter length and how much cache RAM you're willing to throw at the problem. Bear in mind that tape movement in an LTO drive is many times faster than a naïve tape capacity / transfer rate calculation might suggest, as LTO drives require anywhere from a few dozen to a couple hundred head passes to read/write a full tape, and seek to a given logical block number considerably faster than they can read or write. The bigger problem I see with tape library vs. RAID for consumer movie storage is cost. Even if we assume economies of scale would drive the cost of LTO-9 drives down to, say, $1,000, you'd still need to store a lot of movies for the cost of LTO-9 drive + library robotics + tapes to approach the cost of large capacity hard drives + additional cabling and interface electronics, given that $1,000 will buy enough hard drive capacity to (non-redundantly) store over 1,000 50 GB UHD Blu-ray movies. [1] https://en.wikipedia.org/wiki/Linear_Tape_File_System https://en.wikipedia.org/wiki/Linear_Tape_File_System
  - sschueller 3y ago
    Is the cost of LTO due to the technology or the small market and a few large players taking advantage of that?