Digital Storage To Survive a 25-Year Dirt Nap? 1044
AlHunt writes "I've been tasked with finding a way to bury digitally stored photographs in a small underground time capsule to be opened in 25 years. It looks like we'll be using a steel vessel, welded closed. I've thought of CDs, DVDs, a hard drive, or a thumb drive — but they all have drawbacks, not the least of which is outdated technology 25 years from now. Maybe I'll put a CD and a CD-ROM drive in the capsule and hope that the IDE interface is still around in 25 years? Ideas and feedback will be appreciated."
Include a playback device (Score:3, Informative)
Re:How about.... (Score:5, Informative)
Flash memory works by trapping electrons on an insulated gate. Since there is no such thing as a perfect insulator, especially at high integration levels and taking into account quantum effects, those electrons will leech out over time. 25 years is probably more than enough to kill the data on a flash memory chip.
Microfilm (Score:3, Informative)
Screw digital. Get them printed to archival-grade microfilm and store them in a moisture-proof container. Microfilm is designed to last a minimum of 500 years when kept under the proper (and relatively easy to maintain) condition. All you need is a light source and a good lens to view it. Most of the world's better professional archives use a combo of microfilms and digital archiving to keep stuff around... the microfilm guarantees longevity while the digital copy is easy to search and access.
Re:Print them (Score:5, Informative)
I agree completely. Current digital technology is not designed to last for long periods of time untouched. Storage methods evolve, things move around, old hardware fails and new hardware shows up, and data is in a continuous flux. If you shove data into one device and leave it untouched for many years, chances are it will be gone one way or another, since normal storage devices just aren't meant for that kind of use. Flash memory gets erased, hard drives have bearings which stick and die, CDs and DVDs have dyes that can break down over time and aluminum that can oxidize, etc. The proper way of using current storage technology to store data for long amounts of time is to do what we've been doing all along: use normal methods of redundancy (offsite backups, etc), keep the data online, check up on it periodically, and move it over to new storage systems as the old ones become obsolete or break.
If you just want to stick some data in a box for 25 years, printing it out is bound to get you a much higher chance of getting it back. Other means exist of storing data for long periods of time, but consumer digital technology isn't it. Things like laser engraving, coupled with a good reference manual that describes the encoding could work, but these kinds of things are highly specialized and probably not available for a reasonable amount of money. Printing is.
Re:SATA, not IDE (Score:5, Informative)
he can add in a small laptop with a power adapter and a media reader (usb ports, card reader, optical drive, whatever he needs).
that way all he *needs* later is electricity, and id be surprised if the US (or whatever country he is from) has phased out the currently used electrical outlets in 25 years, and even if he did, some electrical tinkerer could power it up anyway.
as for getting them off that device later...thats his problem :)
everything needs to be sealed well, however. double or triple on some water-tight somethingsomething to be safe,
Re:welded closed? (Score:3, Informative)
mod parent up please...
he's spot on. Anybody that has ever welded anything at all will confirm that the inside of whatever you're welding tends to get a little warm, especially if it is a sealed container.
Cold rolling it shut would be a better option, evacuating it and using a simple seal would be even better, especially since that would stop the medium from being attacked by oxygen.
Re:Print them (Score:5, Informative)
I would worry about the media more than the medium (Score:5, Informative)
Generally speaking, you can pull data from media formats (medium) that are 25 years old. If your capsule was to be opened in 50 or 100 years then you'd have a problem, but most media formats that are 25 years old are still readable today. How much effort it would take varies...
If you had a 160/180/320/360KB 5 1/4" floppy disk from 1983, you could even read it by buying an old 5 1/4" drive off eBay, connect it to the same floppy connector that's still in use today, and read the disk directly in Vista. Now, if instead you were trying to read an MFM/RLL hard drive, 8" floppy, magnetic tape, punch card, etc. from that era, then you'd have more of a problem--but it would still be doable.
Also, many companies make specialty products to connect old equipment to new PCs. While I've never seen one, there's probably a company that makes a USB 5 1/4" floppy drive. If push comes-to-shove, you can always buy old equipment to bridge the gap... If I had an MFM/RLL hard drive from 1983, I could always buy an XT or AT from ~1983-1991 (that has an MFM/RLL interface), connect it to a new PC by way of a serial port (well, the new PC will probably have a USB-to-Serial converter) or Ethernet and transfer the data.
Pick a technology that's very well used today and you should be fine reading it 25 years from now. Sure, it'll take some effort & creativity, but it should work. But if you pick a technology that's old by today's standards and you'll have even more trouble reading it in the future...
That being said, I would worry more about the media--whether it will withstand 25 years of isolation, heat expansion/contraction, humidity, etc.
Re:Welded Shut? (Score:5, Informative)
Speaking as someone who welds bike frames, I don't think this is too much of an issue. I can reweld a cracked frame without burning the paint 3 cm away, if I'm using a TIG and doing short welds. Anything they put in there, wrapped in a layer of aluminum foil, should be fine.
Now if someone insists on using an oxyacetylene torch to weld it shut, you have more of a problem, but using a gas torch to weld up a time capsule in 2008 is like using punch tape to store your data in the capsule.
Re:Permanent storage through integrated circuits (Score:5, Informative)
At least Seiko produces serial EEPROMs with > 50yrs data retention, and are rated for high temperatures (125 degrees C).
Those max out at 64k, though.
You'd probably want to use some sort of EM shielding (Faraday cage or similar) in addition to ESD protection, and thermal shielding to keep the temperature on the surface of the die below 125C during welding, and also carefully choose your burying location.
But, yeah, storing thousands (the OP didn't actually say thousands of photos, did he?) of pictures would require thousands of 64k (k-bit, I believe!) of ICs. I can only imagine the programming effort involved; special jigs that house & power hundreds of PROMs per batch write...
Re:SATA, not IDE (Score:3, Informative)
The CD's and DVDs might be fine if you use good archival media, but any reader will likely not work.
-Aaron
CDs (Score:1, Informative)
I have burned CDs that are already 20 years old that work just fine. All of them. I know there were a few burned CD lots that suffered datarot, but I really doubt that CDs are that fragile. I fully expect those CDs to still be valid 5 years from now (even 20 years from now, to be honest).
Still, redundancy is the key, if you have the space. I am certain there will be ways to read ISO or Joliet format CDs 25 years from now. I could read a 5 1/4 floppy disk today, if I had to (those have lasted over 25 years as well).
Re:SATA, not IDE (Score:5, Informative)
Also, after 25 years it is doubtful that any PC or laptop will work. The CMOS battery will likely leak all over the place, and the electrolytic capacitors will not work very well after 25 years (electrolytic capacitors also do poorly when they are unused for long periods of time).
Re:Why not... (Score:3, Informative)
A glass CD master should work fine. A brass CD stamper should, too. The issues then become how to turn those things into a stamped CD.
Learn from the past (Score:4, Informative)
Learn all you can from the (arguably) failure of the BBC Domesday Project [wikipedia.org] and build from there.
My advice for a time capsule is:
It's tough but it just might be doable. Again, the keywords are redundancy and simplicity. If the data is important, make two identical time capsules and store them in geographically different areas (different tectonic plates are safest but this is probably overkill :) ). It's important that the copies of the time capsule be identical so data lost from one can be restored from the other.
Simple: Print it with a laser printer and OCR-B (Score:3, Informative)
Paper has a 100+ years life expectancy. Modern laser prints are not that much worse, at most pages will stick together to some degree, so make sure it is one-sided printing.
As to electronic solutions, data recovery companies should still be able to read CDs in 25 years. But practially no burned CD will survive that long. CVD is not better. As to a thumb-drive, the current data retention times are 10 years, so that is out. HDDs are also out. While they might survive that long, the current visdom is that you should power them up once a year. 3.5" MOD disks are the only viable electronic option, I think. They are used in a lot of medical imaging equipment and in several countries digital x-rays need to be stored for 20 years. MODs can give you that, with a current media life expectanty of >50 years. I would expect that professional data recovery for MODs will be available in 25 years as well.
As to filesystem, go FAT. It is simple and will till be supported, e.g. by Linux.
Quote franklty, the paper solution is best.
Re:SATA, not IDE (Score:5, Informative)
You joke, but I think the best solution would be a macro scale physical recording medium. I wouldn't trust magnetism over 25 years, I wouldn't trust microscopic silicon SRAM for 25 years, and I know for a fact that CDRs deteriorate after about 10 years.
I think the best would be a large stainless steel disk. On the disk, at certain intervals, would be impressions. Each impression would be one of 4 depths of a relatively large difference in height, maybe 2mm. Each depth would represent an octet. Each height difference would be a square milimeter or so. The disk should store about 1000kb for each meter of disk size with both sides used.
Stamped on the middle of the disk would be instructions for reading the disk: What height represents what octet, which way the disk rotates, whether the disk starts at the front or the back. I'd guess the best thing to use would be a laser time domain reflectometer to acquire the data, which could be read one-bit at a time into whatever sort of PC exists in the future. There would be information detailing how to determine the beginning and end of the disk, and at least a reference to whatever graphics standards you use to save the image.
The disk should sit in a vacuum-filled glass case. The glass case should have rubber legs, and the disk itself, being relatively massive, should have rubber legs too. If there's room, a reference telling how to program a reader for the disk and a reader for the graphics standards you've used would be best too.
The pictures should be saved as one massive image on each side of the disk, and that's all that should be on the disk. There shouldn't be a table of contents, only a header and a footer. This will reduce complexity and ensure the person retrieving the data doesn't have to find a 50 year old copy of MS-DOS to run the 25 year old disk.
Doing all this would prevent rust from degenerating the data, it would prevent magnetic fields from disrupting the data, it would prevent subtle chemical changes in the disk from disrupting the data, it would prevent particulate matter from disrupting the data, it would prevent complicated and antiquated drive mechanisms from preventing easy retrieval of the data, it would prevent incompatible future operating systems from causing the data to be unretrievable, it would prevent unknown file formats from causing the data to be unretrievable.
I figure the only thing that could cause real trouble would be if someone made a serious effort to destroy the data surface. Dust could cake the data, but it should be cleanable. As long as there's someone capable of building a reader aparatus (you could include one, but don't depend on it interfacing with anything anybody understands -- let alone working after 25 years), there shouldn't be any hardware issues.
I'm a bit worried about the data density, but that's just the way she goes. A milimeter is a nice macro scale so you don't have to worry about quantum effects damaging the disk, and stainless steel wouldn't rust, but there's a chance of read errors if someone scrapes the disk. Parity mechanisms could be used, but that'd cut the size of the disk substantially in order to provide error correction.
Re:CDs (Score:1, Informative)
Re:SATA, not IDE (Score:3, Informative)
"And evacuate the vessel."
Since the vessel is being welded, one would assume with an inert gas involved (MIG or TIG process) tap that puppy for a pressure fitting and fill it with whatever argon etc. mix is being used. You'll have a better weld by purging it, which is a standard welding process.
It will probably (mostly) bleed off in 25 years, but a vacuum would allow nasties to be sucked IN vs keeping them out.
Paperbak (was Re:SATA, not IDE) (Score:1, Informative)
I was going to suggest Paperbak too.
You can store up to 3 megabytes per page
To be clear, Paperbak encodes 500,000 bytes per page. The 3 megabytes figure was for c code (which is easily compressed) - most image formats are already compressed so you won't get 3 megs worth on a page.
You probably also want to include a human readable description the encoding/decoding algorithm and the source code.
How about making actual photo prints .. (Score:4, Informative)
Re:SATA, not IDE (Score:4, Informative)
Re:SATA, not IDE (Score:5, Informative)
Would storing the laptop in some type of inert gas (nitrogen) help? Hell, should the whole capsule be pumped full of nitrogen to reduce corrosion effects?
The corrosion takes place inside the batteries and capacitors, so no.
Re:CDs (Score:4, Informative)
Just curious - did you work for Sony or Philips? According to wikipedia [wikipedia.org] CDR were first spec'd in 1988, 20 years ago.
As others said, use SCSI if you use hard drives... (Score:4, Informative)
Re:SATA, not IDE (Score:1, Informative)
Microfiche film is currently the best known way to store data for long periods of time.
Adrian Sender.
Cellulose, not silicon (Score:3, Informative)
There are plenty of books out there older than 25 years old... so print out the (binary, hex dump, 1's and 0's) onto paper and store them somewhere where they will survive.
Hell, even that's too tech... engrave the above into stone and make the Long Now Foundation proud.
p.s. Don't mod this funny. I'm not joking. There is no guarantee that any media that plugs into a computer will be easily accessible in 25 years time, but you can damn well guarantee that they will still have scanners, and that their OCR software will have improved since then.
Re:SATA, not IDE (Score:2, Informative)
Not stainless steel.
Even "stainless" steel rusts in the presence of standing water. That's why gold is considered the "gold standard" for artifacts.
If you can't afford gold, a nice ceramic glass compound is probably the next best choice.
Re:SATA, not IDE (Score:5, Informative)
No, it doesn't, not since early mock ups.
Re:You're hosed for actives: Capacitors will fail. (Score:3, Informative)
Aluminum electrolytic capacitors are still often seen on motherboards and cards, but I haven't seen one on a hard drive for well over a decade, and I seriously doubt that they've ever been used in a memory stick. I have some old hard drives from the early 1990's sitting right here. Not a single one has any aluminum electrolytic caps. They all use solid tantalum caps instead, which age much better.
Memory sticks don't need much capacitance, and are usually in thin packages (sometimes as thin as SD cards). It's unlikely that any of them would contain any tantalum caps, much less the much taller aluminum electrolytic caps. More likely just a few small value ceramics, or possibly even just some distributed capacitance layers built-in to the IC substrate, with no discrete capacitors at all.
Re:SATA, not IDE (Score:1, Informative)
> Silica gel is often used in new furniture or goods that are going to be stored for a long time, it soaks up moisture and prevents anything nasty growing inside.
I have quite a bit of experience growing, storing and selling exotic mushrooms, and from that quite a bit of experience using desiccants like sodium silicate (silica gel). While it's pretty effective, it's going to be absolutely useless in an underground box for 25 years unless you're talking about improbably vast quantities of it. If the time-capsule is anything like air-tight enough to prevent the silica gel from reaching its field capacity in a matter of months, the capsule is not going to be absorbing enough moisture from the surrounding environment to raise the relative humidity beyond regular atmospheric levels, so unless you're burying the box in south Florida in the middle of summer, bothering with a desiccant is probably a waste of time.
There's this stuff called "Silica Gel".... (Score:3, Informative)
It comes in little packets which generally have "do not eat" printed on them.
No matter what solution he goes with I hope he stuffs a couple of handfuls of them in there.
Re:Use of USB (Score:3, Informative)
even RS-232 didn't last that long. Tell that to all those microboard programmers. I still see RS-232 ports regularly on new machines and I still have all the cables at home to convert the 9-pin to 25-pin or to a null cable or ... you get the idea.
I don't know how or why you would store any digital data that long. But if you can bribe somebody to put it in the GNU Hurd kernel, it will probably float around within a few years in SVN or FTP or so and within 25 years, maybe you can actually use the kernel as well.
That's all I would do. Keep it online for the next 25 years. The internet isn't going anywhere. I just stored something on a few USB flash drives to keep around for the next 10 years. I know for sure nobody cares about this data within 10 years anyway and most likely it's going to be trashed anyway or it's going to be found by someone that will try to reuse our offices once we vacated them. We just found a bunch of old biology books, floppy drives with stuff on them and brains on formaldehyde in some of the offices upstairs. Nobody cares what they were for, you'll need a historian to find out.