Scientists Suggest Microfiche For Long-Term Archive

It's no secret that we, as an industry, are struggling with the challenge of storing digital data for the long term. And now a pair of academics are suggesting that microfiche--that analog medium we've been using for decades to store old newspapers--could, with 2D barcodes, be the solution to the long-term data storage problem.

One of the bigger problems facing anyone with long-term data storage needs is the requirement to migrate data every few years. Disk systems--whether they have capabilities such as deduplication or Massive Array of Inactive Disks (MAID) to save power by spinning down inactive disk drives--are eventually declared to be at end of life by their vendors. For example, once HugeStor stops servicing your MegaSilo IIIc, you have to migrate the data to a new MegaSilo VIIb or spend long nights worrying about what will happen when the MegaSilo IIIc breaks down.

Tape is somewhat better, but tape formats stay on the market only for so long. While the data on the DLT4000 tapes you sent to Iron Mountain in 1995 is probably still readable, you better migrate it to a new format while you can still get a drive to read the old tapes. Amazingly, consumer optical media like CD and DVD have shown the longest format stability for data, with CD-R now reaching its 20th anniversary of continuous availability.

Microfiche has well-proven data stability. With proper temperature and humidity control, microfiches should remain readable 100 years or longer. The authors tested a 170KB file that contained 191 pages of text. By digitizing and compressing it, the file took up just 12 pages of fiche space. The authors go on to suggest that the resulting fiche could be read by any optical scanner available in the future, eliminating the need for special hardware.

I've always been a sucker for products from the "Gee Mr. Wizard" school. You know, holographic storage, high density optical drives and the like. Reading the technical paper reminds me that I studied chemistry in college, and not just pharmaceuticals and pyrotechnics. So I get a thrill--I know, weird--from reading the academic paper that says that by spreading alternating layers of 2,4,6-trinitrotoluene and 1,3,5-Trinitro-1,3,5-triazacyclohexane each .005 microns thick, a medium is created that can hold 1EB/sq. mm when written by a 10nm laser. I even used to own a Silent 700 portable terminal with bubble memory.These data fiches remind me of British chemical giant ICI's Digital Paper idea of the late 1980s. Digital Paper used pits and dyes like CD-R, but on a flexible polyester substrate that could be spooled into cartridges like tapes. Digital Paper flopped as tape bit density, and speed, grew to the point that they exceeded the capacity of the optical medium.

Therefore, while the idea of storing lots of data on microfiche is cool in a back to the future kind of way, I'm not buying it. First, while dedicated hardware may not be needed, the historian of 2107 who is trying to read one of these new-fangled data fiches will need software that can decode the 2D barcode on the fiche.

Then there's the limited bit density. Storing 170KB in one microfiche page gives us an aerial density of about 1KB/sq. mm. Given that LTO5 tapes store 1.8Mb in that same space, it would take over 20,000 data-fiches to hold the contents of a single LTO-5 tape.

So 2D barcode data fiches go into the trash bin of science fiction storage media with holographic storage and Digital Paper. I wonder what the boffins will come up with next.