Huge Storage Backup at Home - Part III
Raid is NOT Backup
There are a couple of things to think about when setting up home RAID arrays. First of all, I opted to use RAID-5 which uses one drive in the set for recovery data. Don’t make the mistake of thinking that having a RAID array is the same as having a backup. The system is more reliable since it can keep operating if one of your drives fail, but if a second drive fails before you get to it (which can easily happen when you don’t have IT monitoring the systems) your data is potentially gone. In fact with a RAID array it can be much easier to lose your data since any individual file is easily spread out over 5 different disks making it really hard for traditional recovery software to do anything with it.
Unfortunately traditional backup at home is just a pain. I’m a big fan of backing up to hard-disks but most software isn’t designed for these scenarios. Most “backup” software is designed around removable media but I’m really not excited about having to burn 600 DVD-Rs (or even 10!) every week. For example traditional file copy / sync software can easily cause a local delete to destroy the contents of your backup too.
Another problem I hit early on is that one of my drives kept failing. The RAID drivers kept my data intact but this one drive keep going offline forcing array-rebuilds which hurt performance for a period of time. Any good RAID adapter will support rebuild while the drive is still operational. Some cheaper cards might keep your volume offline during the rebuild which can take >24 hours with new huge drives. In any case it turns out the problem was a faulty SATA cable, and I’ve actually hit this more than once. It’s a good idea when you set up your system to stress it for a couple of days with some big file copies and ensure everything is running smoothly before you trust it. If you see a periodic failure with a fairly new drive, try replacing the SATA cable and don’t forget to throw away the old one- I’ve made the mistake of losing track of the bad one in the past.
Managing this much storage can also be a big challenge. I chose to mount my volumes in Windows as paths on my C disk without creating new drive letters- this is similar to the usual Unix/Linux approach and is much more flexible than having to manage drives as “D”, “E”, etc. Think about how you want to expose shares- I’ve got shares for “TV”, “Music”, “Movies”, “Pictures”, “Documents”, “Statements”, and then a couple of more specific shares for specific encoding rates of my music. Another good way to think about your shares is by backup policy. Some of my shares have lots of stuff like old TV programs that I wouldn’t mind so much if I lost them (and besides, they are huge so backing them up would be a pain). My music collection I could reconstruct since it is all legitimate, but it would take me weeks to re-rip everything. Still, it doesn’t change that quickly. My pictures and documents are very precious and I’ve got them set up to backup every night (more on that later).