Large storage solution question
 

I am looking for an external case for 4 SATA HDs that combines them into a RAID (5, probably, but not necessarily) drive, and connects to a computer via an eSata cable. I have seen numerous things like this, except they require a separate eSata/Sata RAID PCI card to be in the computer. Is there anything like this where the RAID controller is built in? I'd like it to be usable with any computer I plug it into.

There are plenty of external RAID enclosures that do not need any software to handle RAID transactions. Though you do need a SATA controller.

http://www.cooldrives.com/fidrsaiihara.html

Thanks. Is there any RAID level that allows for at least 1 disk failure, but where the capacity is not limited by the smallest drive? I can't find one looking through them (wikipedia).

Sorry that is the nature of RAID-1.

Looking at those systems, I've changed my mind. Instead of a giant RAID-5 system, I think what I want to do is have two equal sets of RAID-0'ed drives, one of them being my main drive and one being a (daily/weekly) backup, because this protects against non disk failure data loss and doesn't require me to upgrade every drive (only two of them) when adding space.

Is there a simple hardware solution for this or if is better accomplished with a software solution?

A RAID-0 array will increase your chances of losing data because the data is striped. A RAID-0 does have the limits like RAID-1. A RAID-5 could in fact increase in capacity. When buying a hardware RAID controller or enclosure check if the device can increase the capacity of RAID-5 by just adding a drive to an existing array. Though it is best to back up the data just in case something bad happens.

I realize that RAID-0 will increase the chances of losing data, that's why I am using a backup. RAID-5 won't protect at all from accidental rm -rf, data corruption, etc., will it?

Will ZFS using raidz and snapshots do this? (with the snapshots protecting against accidental rm -rf) Is there anything else like that that will, and will allow the raid to be grown?

ZFS will allow recovering against accidental "rm -rf" assuming a snapshot was made.
Files modified or added between the snapshot and the rm won't be recoverable.

You can expand a ZFS pool without limitations.

Looking for magic bullets? You can get more layers of redundancy, but they all require a combination of money and ingenuity, and they will generally remain vulnerable to a sufficient level of human error -- for example, the user who removes something, can't get things to work, and doesn't report it to you for a week.

So, you could do something like two complete raid5 systems with a backup script copying one to the other during some off hour once a day or once a week. And you could do something like implementing snapshots to catch save changes at multiple different times of day (that's what the NetApp $$$ boxes do). Then figure out the failure scenarios that you are still vulnerable to.

This is why tape backups and tape libraries are widely used in enterprise environments. Then you can recover something from any day of the week going back as far as you care to keep your tape rotation going.

But, since you've started by telling us your solution rather than your environment and problem, it's hard to tell where the appropriate point in this spectrum is for your needs, or how much money and effort makes sense for you.