Thread: Data archiving philosophy questions (LONG)

Originally Posted by sickwookie

If you are going to have three drives it is best to put them in a RAID and use tapes for your offsite instead of having 3 disks in 3 locations. The RAID is going to deal with any single drive failure or corruption and the tapes off site are in case of catastrophe. It is simple, mostly automated, and well proven. The only better solution is to have synced remote systems in place of the tapes which makes it completely automated, and is what you get with an online service.

So, you said in your original post that you preferred KISS. Here it is: RAID 5 with tapes off site. Any additional discussion beyond that is a needless complication to a problem which has already been solved.

EDIT:
And to answer your question on why RAID 5 is better than RAID 1 it is the parity. Every single bit can be verified, corrected, or the entire drive rebuilt without you ever having to lift a finger. This means that even if the probability of a failure or corruption is the same in either array the probability of lost data is drastically reduced in RAID 5.

The question you answered is not the one I asked. And every example you give includes "RAID plus something else" (compared to non-RAID without the extra something).

SIMPLIFY: Imagine I am on a desert island and ONLY have 3 drives. THREE drives... forever. Which is better:

3 single backups (stored at different corners of my desert island)... or one RAID setup (stored near the center of the island).

Any "extras" you can add to RAID... I can also add to a non-RAID setup. But if we're after a true comparison, for the same drive count, I still believe non-RAID is superior.

So, I guess I am not explaining this very well. It happens often sorry for any confusion.

With three drives the best you can do is RAID 1 with an extra off site copy synced regularly. This is basically what you have. I would just go ahead and use RAID 1 instead of a manual mirroring process though. EDIT: This is just my preference. I like to have computers automate as much as possible so there is less for me to screw up.

Additionally, you could significantly improve the robustness of your backups by using all three drives in a RAID 5 configuration and moving to tapes for off site. If it were me that is how I would go.

As far as non-RAID configuration being superior to RAID 5 I am just not sure how you are coming to that conclusion. RAID is very easy to setup, and once setup you gain not only redundancy, but also error correction. This is why it is the standard for any data center or business that needs to ensure the integrity of data.

In your desert island example if you only have three drives to last forever it does not matter how you configure them they will all fail eventually. The only thing you can hope to improve is the time until total failure. This is why I like tape to be included into any backup scheme. By adding tape into that example you can easily lengthen the time until complete failure by an order of magnitude.

Hope that helps you understand why every example I use includes RAID + off site.