Raid-1 is mirroring two disks (if I haven't messed up completely here :p). Does that mean that I could take one disk out and put it into another machine and it would work like a normal disk? Or does it still only work in the raid? I can understand that a Raid-0 of course only works as a whole. But with mirroring I am not so sure how it works...
Of course putting the mirrored disk back in the raid afterwards would be a bad idea, but just taking it out... how would that work? Does raid store any special data on a drive that is needed? or does the raid controller handle everything like that and the harddrives are just used like normal?
Well, the REAL answer is that it depends on the RAID controller. There's Hardware RAID and there's 'Fake RAID' -- which is basically a ROM implementation of software RAID.
Fake Raid will always require the same controller, just as using a linux software raid will require the same operating system (and sometimes the same kernel module version).
Real hardware RAID will SOMETIMES mean that you can swap them between machines, but in some cases you will need to have the same or a similar RAID card. One case where I've seen this is with a RAID5 volume on a Dell PERC/3di card. I haven't ever tried it with RAID1, but I understand it's possible.
Usually, yes, it does.
Sometimes this is true -- either independently from the data on the drive, or in concert with data on the drive. i.e. the drive may have metadata readable only to a certain class of controller that says it was part of a RAID1 pair, and other controllers will ignore that metadata.
This depends entirely on the RAID used. It's entirely possible for different RAID cards and different RAID software implementations to have an on disk format that doesn't let you use a single disk by itself.
However, I know at least Linux's md software RAID lays things out so you can use a single disk outside the RAID.
Yes in most cases this is correct, you can use the disk in a another machine.
It isn't uncommon to use a disk from a mirror set to clone a machine.
Most raid implementations will add a block of data (mdadm calls this a "raid superblock") describing and identifying the array to each disk that makes up the array. It's better to have this information on the disks than on the controller because it allows the array to be moved between controllers and reduces the risk that a disk will accidently be used as part of the wrong array.
This block can be in various locations. For example linux's mdadm offers three choices, beginning end or 4k from beginnning.
If the raid description is at the end of the device and the array is raid1 then you can access the individual drive's seperately since normal partition tables are at the start of the device and garbage at the end doesn't generally matter. This can be both good and bad, it can be useful for booting or data recovery purposes but it can also cause corruption of the array if a partition on one of the component devices is accidently mounted read/write.