The Perils of Long-Term Storage: Page 2

Posted September 22, 2004
By

Henry Newman


(Page 2 of 2)

Continued From Page 1

Issue 3: The format

Tapes have often been written in a strange format. My first job as a CO-OP Student working for the U.S. Department of Housing and Urban Development (HUD) in 1979 was to figure out how to read data written on a tape about seven years earlier. HUD no longer used that tape format and no longer had the tape drive. After a few weeks, I located a drive in the bowels of the Pentagon, but they did not have the software to connect the drive to their current system. After a few more weeks, I found the software from the original hardware vendor, the people in the Pentagon hooked the tape drive up, read the data, and then wrote another tape on another machine and I left with tape that we could read. This was my first foray into the compatibility of formats. I have long since forgotten what the drive type was, but I haven't forgotten the lesson. The same issues apply today.

Tapes and drives are not backward-compatible for very long. Five years seems to be about it. Add to this the problem of the hardware interface: in a short time in history, we have seen IPI, SCSI-WD, SCSI-FW, Ultra-SCSI, Ultra160-SCSI, FC-AL, FC-Fabric 1 Gb, FC-Fabric 2 Gb, and soon, FC-Fabric 4 Gb. This does not bode well for future tape drives, interfaces and backward compatibility. What OS vendor is going to maintain drivers for tape drives that are no longer used?

Issue 4: Using the data

Formats for most applications change over time. Try using a document created in Word 1.0 or a very old version of WordPerfect or Excel or any other application. If you digitize movies, will you find the correct MPEG decoder in 20 years? The safest format for movies, from what I have been told by people in the business, is copying to another film. Adobe PDF is guaranteed to be a compatible format for 30 years, but other formats are not. Migrating data from one tape format to another gives you the opportunity to update the application format, which is likely a good idea given how often formats change.

Conclusions

Tapes have some significant advantages over disk for long-term storage, but with those advantages comes the requirement for migration. Long-term, I am not sure that current disk interfaces won't have the same migration problem. Just 10 years ago, we were running RAID devices using SCSI-FW. You can probably find a device driver in most vendors' operating systems for that RAID, but will you have a file system that is compatible that you can mount? Whether it is tape or disk, you are going to have to migrate something.

Here are some steps you can take to prepare yourself for eventual migration:

  • Find out how long the technology you are using will be supported.
  • Understand the costs — including time and personnel needs — of what you will need to migrate.
  • Develop a good understanding of the technologies that are available and their planned migration paths and upgrade paths.

You need to determine what technologies meet your needs for cost, longevity and reliability, because without this information you cannot make the best choices for your organization. Vendors will try to sell you what they have tomorrow for these types of problems; make sure you get the vendor's roadmaps from previous years to make sure that they have met their promises.

You need to view migration as a fact of life, for tape or for disk. With the coming T10 Object Storage Device (OSD) standard, file system migration is going to be a must. I still think that tape is cheaper than disk for long-term storage, even given the migration issues. But whatever your choice for long-term storage, migration must be part of the plan.

See all stories by Henry Newman


Page 2 of 2

Previous Page
1 2
 





0 Comments (click to add your comment)
Comment and Contribute

 


(Maximum characters: 1200). You have characters left.