Backup Best Practices

Options for performing the all important task

[hannamyluv]

I am not a very hardware savvy person so up until now, I have been backing up important site files by using zip and a 1G thumb drive. This has now become way too tedious. Too many files to do this.

What is the best way (or some different ways) that I can back up my files? How often do you guys do this? Any horror stories to further encourage me to regularly backup my files? :)

[bill]

I'm a big proponent of imaging my drives. You can do this with a number of software packages out there. Symantec's Ghost 9 is what I'm currently using and it has saved me on multiple occasions. Just schedule backups to run when you're not at the PC and let the software do the rest. If ever you need to get back to an exact previous version of your machine all you have to do is restore. (You can restore individual files or entire partitions.)

[JAB Creations]

I don't know squat about raid but there is a raid number where you buy twin hard drives (has to be exactly the same model numbers) and while you only see one drive, the other is always copying (be it good/bad) what the other is doing. It's useful if you know what you're doing in Windows and don't enable Java for viruses or download stuff from file sharing (spyware).

I suggest reading Wiki's raid page...
[en.wikipedia.org...]

[bill]

RAID makes for lousy backups. It's good at hardware redundancy and data mirroring though.

If I have a set of backups I can restore my system to a point in time when it was working. Or I can restore corrupted files. In RAID you'll just have multiple versions of a corrupted system/files. RAID is not a substitute for backups.

[2by4]

Do both if you're really into it, the raid for never losing any data through disk failure, which is the most likely failure you will experience on the hardware side, and the backup for replacing messed up, lost, misplaced etc files.

But the raid adds a little cost to the box.

Diskimaging I'd like but it kind of assumes you're running a single drive, doesn't really do all that much good for multiple hard drives from what I can see, it's easier just to create one backup file, I actually installed ghost 9 before I realized that of course it's just going to ghost a disk onto another disk, fine for a simple machine like a workstation or something.

[bill]

I'd recommend different products if you're looking at server backup. They do have imaging products for those sort of setups, but it becomes a bit more expensive.

You can image multiple drives just fine with Ghost. It's not what you'd want to use in a RAID config though...

[wheel]

There's one word you need to remember when it comes to keeping backups current:
Automated

If your backups aren't automated you are going to neglect them at some point. Maybe today, maybe tomorrow, maybe in six months. But at some point you're going to realize it's been two months since your last backup.

I keep a backup server in the basement with a couple of scsi drives and a dvd burner. All desktops and my webserver get backed up every night to it.

Also each night, the backups are zipped and copied to a second drive on the machine (filename is the date). This gives me a complete backup of all my machines for every day.

Then once a week or so I burn a dvd of a few of those days and delete them.

I also raid my hard drives on my webserver, but that's more for redundancy than backup. From what little I've read, raid has it's own problems so I do it, but don't count on it.

I also have a spare hard drive on my webserver where I backup the webserver to each evening, just to have an onsite backup. (hard drives are cheap).

It took a bit to set up this system but I no longer worry about my data backups or archives. The only thing I've got to do is burn a dvd on Friday afternoon and if I miss a day it's no big deal.

It's a cheap, effective solution as far as I'm concerned. A cheap machine in the basement with a burner and a couple of hard drives doesn't cost much these days. It certainly doesn't need to be top of the line equipment.

[2by4]

wheel, that's a pretty good setup, gives me some ideas, if you go with standard commercial backup solutions, they get very expensive once you get over a few hundred gigabytes, but a cheap box with very large hard drives, attached to the network would only cost about 300/400 or so if you use an older system, and could just be turned off, that's very cost effective.

At current prices, you could store terabytes of backups for very little money, a tiny fraction of the cost of commercial backup solutions.

For offsite data backup you can just use removable hard drives.

[MrSchmidt]

There's a free little app I use called Easy Backitup
[rdcomp.net...]

Just select the files you want backed-up, set the backup schedule, point it to your thumb drive and leave it alone. Just leave a thumb drive plugged in all the time or backup to a second HD if you have one.

ED

[bill]

wheel what sort of software are you using for the backups? Are these all Windows machines? That does sound like a great setup.

[trillianjedi]

I operate a similar system to Wheel on my linux boxes - but I use FTP to move the files around.

I use a shell script to tar up the entire home directory, gzip it, then FTP it over to a remote machine with a datetime stamp in the filename.

CRON calls that for me every day at 3am.

[wheel]

I installed Mandrake linux (which is free) on my machines that hold the backup. I use rsync as the software that does the backup. Rsync is a linux utility that allows you to only backup/transfer changed files. So the first backup is big, the others less so. Then I exclude from the rsync command my log files and any other garbage I don't want transferred.

If you're doing this, you need to make sure that the machine initiating the transfer can login to the backup machine without using a password. Not the greatest of security procedures, but I'm not too worried about it as I assume the initiating machine is secure :).

Other than the cost of the computer and the cost of a static IP address on my home connection, the setup is basically free. But more importantly it's both automated and allows me to maintain control over my data.

This is the batch file on my webserver that backs up my home directory, my mysql files, and my /etc config directory. Runs at 3 each morning.

rsync -e ssh -azv /home my.home.ip.address:/backup/webserverbackup --exclude "/path/to/logs" --delete-excluded

rsync -e ssh -azv /var/lib/mysql my.home.ip.address:/backup/webserverbackup --delete-excluded

rsync -e ssh -azv /etc my.home.ip.address:/backup/webserverbackup --delete-excluded

Then I run this at 8am on my home backup machine. It zips the file, gives it a name that includes todays date, and moves it to a second disk drive on the same machine.

tar -czf $(date +%Y%m%d)webserver-backup.tgz directory-that-contains-the-backup

mv $(date +%Y%m%d)webserver-backup.tgz /backup2

edited to add: My setup looks a lot like trillian's, except I'm using ssh instead of ftp. The transfer is gzipped and encrypted, though the encryption doesn't do much for me other than as general principles I don't run ftp. For most folks the ftp might be better - maybe you don't need to worry about login?).