The reason I ask is that the SansDigital enclosures I got don't have any internal logic, other than the port multiplier. To my knowledge, they don't manage the RAID volume internally. The logic is done on the controller. This may be a different case if yours is USB only. Even if this is the case, I can tell you the router doesn't have any way to manage the device, so if you lose a drive, you won't be alerted unless it has an audible alarm or something.
It looks like they have two different kinds - one type that uses the port multiplier, the other uses an internal (to the enclosure) controller. If he's got this one
then it apparently manages the RAID volume itself.
1. In general, best practice is to use all the same drive models for best performance. Since these drives are all from the same drive manufacturer, and all 7,200 RPM, I don't see a big issue. It should work fine. I don't think the cache will make that much of a difference.
There's also a school of thought that says you should use *different* brands/models, since with a single type of drive there could be a design or firmware defect that causes multiple drives to fail at nearly the same time, killing the entire array.
2. Since the RAID itself is coming from the enclosure, it is easily portable. My experience with routers, though, is that they are finicky on what storage they will see (even with one disk). So while the RAID is portable, the router may not see it.
Another potential downside of connecting it to the router is that the router will likely become a performance bottleneck.
Two final thoughts:
RAID-5 suffers from poor performance when you lose a drive.
Yeah, but if you lose a drive you should be replacing it and rebuilding the array ASAP so the poor performance is only temporary.
A potentially bigger concern is that depending on how they've implemented the RAID controller in the enclosure RAID-5 writes could be painfully slow even on a healthy array. I would be a little surprised if a $150 RAID enclosure has hardware accelerated RAID-5 parity calculations; typically it'll cost you at least that much for just an internal controller card with hardware parity.
Another issue with some RAID-5 implementations is that after a drive failure, if you get a read error on one of the remaining drives during the rebuild or accidentally swap out the wrong drive, you can lose the entire contents of the array.
RAID-1 avoids the performance penalty for degraded arrays, the (possible) performance penalty for RAID-5 writes, and the risk of killing the array from an incorrect drive swap. It might be worth considered doing a pair of RAID-1 arrays instead of a single RAID-5.
RAID is no substitute for backups. If the data is important, back it up!