Don't backup? Start here.

DougLite

Supreme [H]ardness
Joined
Jan 3, 2005
Messages
4,764
For files such as financial records, pictures, home movies, etc, backups are vital. They are the only way to protect this data against loss. You can always reinstall your operating systems, and you have installation media for your programs. However, you must take care of your irreplacable personal data. There are a variety of ways to do so. They vary in complexity and effectiveness, and a backup model that works for one person may not work for you. Also, a backup strategy does nothing if you don't stick with it and keep it up to date.

When planning backups, there are six main causes of data loss you need to consider. We will evaluate each backup strategy's sbility to prevent each cause of data loss, as well as its ease of use/simplicity. Simplicity for each method will be arbitraliy rated on scale of 1 to 10, with 10 being easiest. Capacity and cost are also concerns, and will be ranked from 1 to 10 as well, with 10 offering the cheapest cost per unit of storage and highest raw storage capacity.

Hard disk failure
Hardware failure (PSU blow up, etc)
Corruption
User/Error mistakes
Malicious activity (virus, hacking, etc)
Natural disaster (flood, fire, etc)

RAID:
Simplicity: 10
Capacity: 10
Strong vs: Hard disk failure
Useless against: Everything else :eek:
RAID is very popular and simple. Once you setup a RAID array, you can forget about it as long as the drives in the array remain operational, as the RAID BIOS automatically handles keeping the data on all of the disks. The RAID BIOS also rebuilds the array once you supply a replacement for a failed member.

However, RAID has some key weaknesses because of the simplicity. A RAID array has one file system, making it vulnerable to corruption, user error, malicous activity, natural disaster, and any hardware failure of a component other than hard drives. RAID protects uptime, not data. RAID will not protect your precious data from a virus or you accidentally deleting it. Also, your RAID will die in a natural disaster.

External hard drive:
Simplicity: 9
Capacity: 9
Strong vs: HDD failure, hardware failure, user error
Weak vs: Natural disaster, corruption, malicious activity

External hard drives, especially if you make one yourself, provide gobs of cheap capacity and the ease of drag and drop or even automated scheduled backup. If your box goes down, restoring from an external HDD is pretty quick. If you accidentally delete a file or save an unintended change to it on your system, resotring from an external HDD will allow you to go back in time.

However, an external hard drive only provides one layer of protection. If you copy a screwed up file to the external drive, you can't use it to recover from corruption or user error later on. Also, an external drive is vulnerable to malicous activity and corruption while it is connected to your system, so only have it connected and powered up when you are actively performing backup/restore procedures. An external hard drive will also most likely go down in a disaster such as a fire or flood if you store near the system you are backing up - consider offsite storage, but that makes it vulnerable to malicous activity (theft) unless you are sure that offsite location is secure.

Backup to another system:
Simplicty: 7
Capacity: 6
Strong vs: Corruption, user error, malicious activity, hardware failure
Weak vs: Natural disaster

Backup to a separate computer has a lot of advantages. If is pretty unlikely that two boxes would be hit by corruption or hardware failure at the same. You can simply drag and drop to a network share or schedule backups. A second system has virtually unlimited capacity, but you have to go to the trouble of keeping a second system running. This method is not quite as strong against a virus as it is aginst other causes of loss. Also, if both systems holding the data are in the same building/area, they are still quite vulnerable to being wiped out in a single natural disaster.

CD/DVD:
Simplicity: 4
Capacity: 3
Strong vs: Everything

Backing up to CD or DVD, when done right, can protect you against virtually every conceivable cause of data loss. It is pretty easy to have multiple copies of critical data stored in multiple locations. Also, CD-Rs/DVD-Rs are write once media, making them extremely resistant to malicious activity, even in storage. However, backing up to CD/DVD falters where high capacity is required, as swapping discs in and out quickly becomes tedious. Hopefully, high capacity DVD formats (Blu-Ray/HD DVD) will defeat this limitation.

Tape:
Simplicity: 3
Capacity: 4
Strong vs: Everything

Tape is like CD/DVD, it can protect against pretty much any cause of data loss if you have a good backup schedule and offsite storage. It also available in capacites measured in the tens of gigabytes, and is about the only choice right now for large storage of absolutely vital data. Tape requires substantial startup investment, but there is a reason businesses use it, despite its tedious nature.
 
Snapshotting
Simplicity: 4
Capacity: 10

Strong vs:HDD failure, hardware failure, user error, malicious activity, some kinds of corruption
Weak vs: Natural disaster

There's another option that wasn't mentioned - snapshotting. You could keep your documents in a folder, and have your server grab dated backup copies of that folder with little more data size cost than just keeping one copy. `rsync`, a Linux utility, has a method of doing this nicely, and there's a Windows version as well. Sun toasters also have a way to do this as well. This in combination with raid levels (10 or 5) provides a good, fast, easy method to back up data. It usually requires a dedicated machine, but many [H]ers have this requirement met.

This strategy reduces the manual labor of tape swapping, and if a virus destroys your primary copy, you can still retrieve a copy of your files from a recent date. It's also useful for businesses; you can retrieve any document from any date, and when the statute of limitations is up, you can have the backups from that date automatically destroyed. Thus, in a high-liability situation, you will have all the files you need and none you don't want readily available. For home users, this is also a good idea; you can store all your documents or your mp3 collection or whatever in a snapshotted directory, and whatever changes you make to the data set, only changed files will be stored. I kept backups of 40gb of data on a 60gb disk for about 2 years, and the disk didn't fill. Since the backups happen once a night, most temporary files and whatnot get cleared out in that amount of time, and most files just sit there taking up space 99% of the time.

It is also possible to tunnel all the traffic to this server over a public network (the Internet, for example) via SSL. This allows easy offsite nightly backups if you have a colo or something that you need to back up.

In any case, something to mention. I'm a big fan of it ;)

 
Back
Top