Backup, Backup, Backup

A friend of mine tells a sad tale of his wife's web business.

He is a very competent Windows system administrator, able to deal with both the hardware and software. He shops at garage sales for bargains and makes a reasonable living off referbishing laptops and such. She makes her income by doing some quite extraordinary graphic web site setups.

She used a 500 Gig external backup drive to store her archived works. Needing to refer to one of them, she hooked in the backup drive and poof... it died.

He took this external USB drive and tried to get it to work on another machine - poof, the USB slot no longer worked!

He took the drive apart and pulled out the raw SATA drive, and mounted it in a machine with a PCI-based SATA card. It blew up the card.

Suffice it to say, the drive is dead - badly dead - disasterously dead.

There isn't a backup of this "backup".

In fact, this "backup" isn't really a backup - it was an offline "archive" drive and should itself have been backed up because there was no other copy of much of the content of the drive.

Let this be a lesson to you... the real meaning of "backup" is "redundant" - which implies at least 2 copies exist!!!!

In my world I keep virtually all of my data "online" because its growth has fairly closely followed that of the growth of hard drives as my systems evolve.

The evolution of the last 4 systems has been from RAID 5 (several disks with a "parity" disk plus a spare) to RAID 1 (mirrored drives, also with a spare)

Almost all of this is also backed up by offline drives of various sizes; mostly 300-320 Gigs. Some of it is also on DVD but I have started to read them back onto hard drives as they have their own problems over time.

Hard disk drives so far are the most reliable offline storage for large amounts of data. By using a mix of manufacturers and sizes it is most likely that specific problems will be side-stepped over time. 

As an example (only an example - the names have been changed to protect the guilty)

My digital images have been backed up first to CD, then more recently to DVD, and still more recently to a couple of hard drives totaling about 400 Gigs. The transfer from CD turned up two of the 40+ that were bad, but since I had duplicates on three different CD brands, I lost nothing.

The transfer from DVD to hard drive went flawlessly.

The hard drives are from two different manufacturers and are of two different "generations" (100 Gig and 200+ Gig) and when I recently went to read them I found that some of one manufacturer's drives took longer to read than I had hoped. It turned out that the drives exhibited a thing called "stiction" which is to say that their motors didn't come up to speed immediately.

In the above case I lost nothing - but in the past I have lost many things due to simply not having adequate ongoing copies in newer technology. This is in general called "bit rot" and is an ongoing problem for everyone.

What you should take away from this article is that no matter whether some particular medium worked flawlessly when you copied your information to it, you should be thinking ahead to when you actually expect to read it back.

I am also faced with a number of tapes from three different generations of drives and two different styles. No player for any of them is available new. I have several samples of drives that should read the tapes but at this point I have literally boxes of tapes (from 60 Megs to 2 Gigs each) that I cannot read.

• CD and DVD media ages and deteriorates over time - as little as a couple of years in some cases
• Hard disk drives have motors - and a motor that has run fine for many hours can suffer from a seized bearing after it has lain unused for some time (and cooled down) - lubricants have a "shelf life"
• technologies change over time and in come cases have changed literally overnight. The DAT (Digital Audio Tape) drives our company used in the early 1990s had proprietary compression - and when the company went out of business, we lost access to new drives to read the old tapes.
• Open standards are best. The proprietary compression scheme died with the vendor. An open scheme might (would) have lived on since I might have been able to adapt a bare drive to the task with my own driver software if I'd had access to the algorithm.

Take heed - your data will disappear if you don't

richard