Highly Reliable Systems: Removable Disk Backup & Recovery


Why Tape Backup Doesn’t Work for Data Storage

By Darren McBride

Why Tape Backup Doesn’t Work for Data Storage

By Darren McBride, BSEE

Through the course of a long career I have worked with many types of tape backup appliances,¬†and consider myself a reasonably knowledgeable user of these systems. I have come to believe that tape does not belong in any serious backup strategy if any other data storage alternative can be found. Admittedly, in some cases the cost of tape per Gigabyte of storage can’t be beat, but the cost savings are often artificial when you consider the massive amount of human effort wasted struggling with a tape backup or restore that doesn’t work.

Consider the following general problems with tape backup and tape drives:

  • Tape backup media is slow and often difficult to restore from. Tape is not “random access” media. This means to retrieve a particular file requires fast forwarding to a certain spot on the tape. This often results in substantially slower data recovery times than if you wanted to retrieve a file from a random access media such as a hard drive (or even a slow floppy drive). Often the speed is so slow it is more cost effective to redo a file or retrieve it in any number of other ways leaving retrieval from tape as a “last resort”. In an attempt to compensate for this shortcoming, tape backup software vendors such as Backup Exec often create “catalogs” on the local hard disk to speed forwarding or rewinding the tape to the approximate proper location. If these catalogs get lost or corrupt, data retrieval is even SLOWER than usual.
  • Tape media is extremely sensitive to dust and dirt commonly found in today’s non-filtered server rooms. Businesses in the concrete, mining, construction and other “dirty” industries frequently burn through tape drives and start getting errors in less than 4 months. Even relatively “clean” industries and office environments often have problems associated with paper dust or other air-borne contaminants.
  • Tape media is basically magnetic particles on a flimsy polyester base film. The base film is subject to stretching, expansion and breakage. With the tight tolerances required of the read/write heads a normal tape can stretch and easily cause errors when reading or writing data. Despite what the tape vendors say, the higher the density and speed of the tape the more sensitive the tape will be.
  • Tape is exposed to air, and the moisture and contaminants in the air. Tapes are not hermetically sealed as is the case with hard drives. Thus, there is no inherent protection for the problems discussed above. Even mold and mildew are possibilities in moist climates.
  • A tape cartridge exposed to any serious high temperature is in danger of melting. There are many cases of hard drives surviving fires, and having data recovered (see these examples). Such stories are far more rare with tape.
  • In the real world, I have observed that good tape cartridges are reliable for no more than about 50 uses. I should note that this flies in the face what tape manufacturers will claim in their spec sheets. They often specify “duty cycles” of 100’s of thousands of uses. Although I’m sure some tape “expert” will contest a low usage number like 50, I found it extremely interesting on a recent visit to a TV station to notice “tick marks” on the DLT tapes they use for storing commercials. When I asked about it they said they see errors if a tape is re-used more than 50 times so they started tracking usage and archive the tapes after that many uses. This exactly coincides with my own observations in the computer world where backup tape should be rotated out after about one year of weekly use (about 50 uses). Technologies such as Travan or DAT won’t even last that long.
  • Different software uses different formats to write to tape media. So, for example, a tape written to using Computer Associates Brightstor cannot be expected to be read using even the same tape drive with Veritas Backup Exec software. The lack of format standardization has been a constant theme in the tape world.
  • Tapes written on one drive often fail to read on another drive (even if the second drive is the exact same model and manufacturer). It is often explained this is due to differences in “head alignment”. This problem has been around since tapes were invented and it is completely unacceptable that your data be vulnerable to the failure of the specific tape drive it was written on.
  • Tapes written on higher capacity drives (for example, 40GB) cannot be read on lower capacity drives (for example, a 20GB drive). Despite tape vendor claims, tapes written on lower capacity drives often STILL cannot be read on higher capacity drives due to tape alignment problems discussed earlier.
  • Tape shelf-life is a guess. Although there have been numerous studies about tape longevity and stability that have produced valuable information, such as the work conducted by the National Media Lab in the mid-1990s, an accelerated aging test that produces meaningful quantitative data about magnetic media longevity does not exist. Some experts state that generally magnetic tape “lasts” anywhere from ten to sixty years. They tell us that “The principal means to prolonging tape life is to maintain an appropriate player, to keep original materials in stable, cool and dry storage conditions and to strictly limit the use of original materials” (See reference to quoted text below) But how can this be done in a “real world” network backup environment? It is simply unrealistic.
  • Tape drives Fail – We all know electronics fail. But it seems to me that whether it’s the moving parts in a drive or the extremely sensitive components, the number of returns on backup tape drives greatly exceed similarly complex electronics systems. I have no data to back this observation up, as it is an opinion based on personal experience in the computer and server backup industry.
  • Consider the following list of Do’s and Don’ts of handling video tape from the web link above (Although this discussion is for video tape realize the technologies and techniques are similar). Does all the required “special care” listed here sound like the kind of robust technology that you want to trust your data to?

DO

  • Learn and use correct procedures for operating equipment.
  • Handle backup tapes gently.
  • Keep tape backups in protective cases when not in use.
  • Keep backup tapes vertical when not in use.
  • Make sure machine alignment is correct before use.
  • Clean tapes before playback if they show any evidence of dirt or contamination.
  • Ensure that the tape is properly seated in the machine before use.
  • Wind tape at low speed (library wind) entirely onto one reel after use.
  • Secure tape ends on open reel tapes.
  • Package tapes adequately for protection before shipment or transport.
  • Use only new tape when recording a tape for long-term storage.
  • Activate the Record Protection feature of all master cassettes immediately after they have been recorded.
  • Inspect tapes for damage or contamination before use.
  • Seek experienced help as soon as possible in the case of a disaster.
  • Protect both tapes and machinery from dust and debris.
  • Keep tapes in a stable an environment.
  • Acclimatize tapes before use if they are hot or cold.
  • Store tapes in a cool and dry place; see ISO 18923.

DO NOT

  • Touch tape surfaces with bare hands.
  • Put pressure on reel flanges.
  • Stack or place objects on top of unprotected tapes.
  • Force tapes into cases or machines.
  • Drop or throw tapes.
  • Splice any portion of a tape.
  • Place tapes on or near sources of magnetic fields.
  • Play or spool tapes that are dirty, contaminated or wet.
  • Play or spool tapes on a dirty, misaligned or malfunctioning machine.
  • Store tapes in an area subject to dampness or possible pipe leaks (e.g., basements).
  • Expose tapes to food or beverages.
  • Expose tapes to temperature extremes.
  • Expose tapes to UV radiation, including the sun, for extended periods.
  • Attempt to clean tapes contaminated with adhesives, fungus or unknown substances unless you have the necessary training or experience.
  • Expose tapes to high power biological decontamination scanners. High levels of radiation can produce sufficient heat capable of melting or deforming tapes or their plastic containers.

In closing I have to say I believe Tape vendors in general have poor technical support, poor ethics, and won’t admit the technology is problematic. As an example of poor ethics I site the common industry practice of quoting compressed capacities and speeds for tape. In the real world the compressibility of data is an unknown and thus should not be a prominent part of the marketing of the capacity of a tape.

As another example I would point to the poor technical support common in the industry which exacerbates the number of hours spent troubleshooting problems by at least 2 or 3 times as they put us through paces with bad advice that has nothing to do with the basic problem. They have us check the SCSI terminators, flash the Bios of the drive, clean the tape heads and try new backup tapes. By this time we’re so frustrated they hope we will figure WE are the ones doing something wrong.

If you have a choice, I recommend you just say NO to tape backup!

Darren McBride

About Darren McBride

CEO, Highly Reliable Systems, Inc. View all posts by Darren McBride →


What do you think?

Your email address will not be published. Required fields are marked *