Do you have a way to determine the number of backups you should keep of a bit of data?

[mikeymikec]

I have two backup drives, each with an 'everything' backup set of my data that I wouldn't want to lose. It's not quite everything but I wouldn't be crying about what was missing should the worst happen.
I also have a drive for storing as many backup sets as possible of particularly important bits of data such as my main documents folder, Firefox and Thunderbird profiles, etc., and each backup set of each bit of data is datestamped in the filename.

When catching up with my backups today, I decided to remove a particular bit of data from my 'multi backup set': Due to my line of work I store locally the manuals for say motherboards I've used. While I don't expect these nor their parent companies to just disappear from the Internet at any point soon, companies can make boneheaded decisions and companies do sometimes go under, however I thought that having three copies of that particular bit of data was more than enough and since I don't go deleting or editing data in that folder very much if ever, the likelihood of corruption is minimal. However I obviously decided at some point that my 'manuals' folder was important enough to warrant the extra protection (perhaps because at the time I decided it was a fair bit smaller than it is now?), so I'm wondering if I'm looking at this situation at enough angles to make a sound decision.

So my question is basically how do you decide that a particular bit of data is important enough to make the grade of storing lots more backup sets, do you have a rule of thumb that you go by?

 

[mindless1]

Value of data (in money/work-hours or irreplaceability/sentimentality) vs cost to store X # of copies, including consideration of the effort needed to maintain X # of backup sets.

As far as motherboard info/manuals, and drivers (that I store by chipset instead of per motherboard so I don't have multiple copies of each in the parent directory), those along with any other compute-related hardware pics/docs/drivers/etc are stored in 3 locations, two on a fileserver on different drives, and one on an offline external HDD used to backup the fileserver. I'd guess that's around 20GB x 3.

I'll probably soon start storing all that on a USB flash drive too. I already have all that for the motherboards and other hardware I still own and use, on a flash drive. Oh wait, I probably have a 4th copy on a HDD I pulled out of the fileserver when I swapped in a newer larger one a few years back, but who knows if that HDD still works after sitting around all this time.

I never assume that at some later date I'll be able to get something off the internet even though that has been mostly true, with exceptions like Abit going under when my main rig had an Abit board in it, but they kept their site up for what, 3 years after that?

 

[Rifter]

3 copies with one being cold and offsite should do it. I keep my offiste backups at my bros house, and have one copy backed up local as well.

 

[Captante]

3 copies with one being cold and offsite should do it. I keep my offiste backups at my bros house, and have one copy backed up local as well.

Old IT saying: "If your data doesn't in at least 3 places, it doesn't exist at all"

 

[badbanana]

I have two backup drives, each with an 'everything' backup set of my data that I wouldn't want to lose. It's not quite everything but I wouldn't be crying about what was missing should the worst happen.
I also have a drive for storing as many backup sets as possible of particularly important bits of data such as my main documents folder, Firefox and Thunderbird profiles, etc., and each backup set of each bit of data is datestamped in the filename.

When catching up with my backups today, I decided to remove a particular bit of data from my 'multi backup set': Due to my line of work I store locally the manuals for say motherboards I've used. While I don't expect these nor their parent companies to just disappear from the Internet at any point soon, companies can make boneheaded decisions and companies do sometimes go under, however I thought that having three copies of that particular bit of data was more than enough and since I don't go deleting or editing data in that folder very much if ever, the likelihood of corruption is minimal. However I obviously decided at some point that my 'manuals' folder was important enough to warrant the extra protection (perhaps because at the time I decided it was a fair bit smaller than it is now?), so I'm wondering if I'm looking at this situation at enough angles to make a sound decision.

So my question is basically how do you decide that a particular bit of data is important enough to make the grade of storing lots more backup sets, do you have a rule of thumb that you go by?

easy peasy. if you can't work or live without those data, then it's worth the extra protection.

and don't backup everything! LOL. keep your data using only a singler folder, say "My Documents", and backup only that folder. if your intent is to do a recovery after a failed drive, then get an drive imaging solution, create an image of your drive, store that image in three places easily accessible on an emergency.

 

[Red Squirrel]

It's more of a personal thing than technical but basically ask yourself what you would do if that data was gone. Irreplaceable things like memories (ex: pictures) or important documents that would be hard to reproduce from scratch (ex: resume or other such documents) should be backed up and archived and at least one copy should be offline. Stuff that requires tons of work (like code) should be backed up very well too, but it's "slightly" less critical than the memories etc that can't be recreated. Stuff that you just hoard or don't REALLY need like downloaded content is probably fine on raid and a simple rsync backup to another raid.

Either way, I think both raid AND backups is very important. Raid will protect you from HDD failures without causing much interruption. Backups will protect you from accidentally deleted files or if you want to go back to an older version, or a total raid failure etc.

For important stuff where I want some level of versioning I usually have rsync jobs that backup to a month/dayofweek folder. This gives me 7 backups per month that I can go back to. There's also rdiffbackup that is probably better I just never got around to playing with it, it's more involved.

 

[SamirD]

Every bit of data should have at least 3 copies since bit rot can set in and it would be nice to know which file is unaffected--you need at least 3 copies of the data to deduce this. NOw, that being said, if 2 of your sets of a file have bit rot, you still won't know, so you need a 4th backup.

And as far as what to backup? Everything. Storage is so dirt cheap today that you can throw away a drive and lose less time than going through the drive to 'purge' files. Besides, who's to say what you will or won't need until you need it? Keep it all, multiple copies, and if you ever need it you can find it.

Most importantly, figure out a methodology by which you can save and find things infinitely. There's nothing worse than knowing you have the data, but have no idea how to find it due to everything being a mess.