rules of backup

Think like a pessimist, hope like an optimist

The amount of money to budget for backup costs should equal how much it cost to originally create the data (man hours, customer interaction, equipment). If the data can not be reconstructed, then what is the cost of loss? Most businesses are built on information. Backup is about that information being accessible.

• • If someone else is in charge of data you need, then you will lose access to it

    • through their incompetence, politics, miscommunication

    • solution -> If you need the data, then you need control of the data.

  • If you have three backups, then

    • the source machine will crash

    • the first backup will be corrupt

    • the second backup won't actually have been copying

    • the third backup will be lost

    • solution -> use md5 sums, verify that the backup is valid by testing it, and never rely backups to work

  • if you have all your backups on-site, then Godzilla* will attack

    • * Godzilla = fire, bomb, anthrax, theft, nuclear attack, chemical/biological attack

    • solution -> copies should be stored off-site

  • if you have backup, it will be stolen/lost

    • solution -> use encryption

  • if n technicians know how to access and restore from backup, then they will all be on vacation, sick, or hit by bus

    • solution -> if you need data, then you need to both have access AND know how to restore

  • RAID is NOT backup

    • rm -rf / * affects all the drives in the RAID stack

  • if doing nightly backups to an external disk

    • "cp -r" (recursive) is not the same as "cp -r -u" (update) is not the same as "rsync -av"

    • use rolling backups

  • if all backup is on one media, then that media type will stop being made in less then 6 months

    • solution -> diversify your storage media types

  • Corollary: if your backups can only be read by one software or OS, that OS/software will be extinct

    • solution -> opensource virtualization software (virtualbox) of the OS with software for reading the backed up data

  • your storage environment control system will fail

    • heat, humidity, sunlight

  • when you restore a backup, it will be under time constraint

    • solution -> practice restoring

  • you need something to backup to

    • this is part of disaster recovery: a backup media is useless without something to restore it to.

    • solution -> redundant, off-site servers that can be accessed. Off-site laptops available for use by workers in the event of primary site failure

In conclusion, your solution needs to address these concerns.