Even given what one of the other answers here laid out, namely that ZFS only works with actual used blocks and not empty space, yes, it is still dangerous to make a large RAIDZ1 vdev. Most pools end up at least 30-50% utilized, many go right up to the recommended maximum of 80% (some go past it, I highly recommend you do not do that at all, for performance reasons), so that ZFS deals only with used blocks is not a huge win. Also, some of the other answers make it sound like a bad read is what causes the problem. This is not so. A bit rot inside a block is not what's going to screw you here, usually, it's another disk just flat out going bad while the resilver from the first disk going bad is still going on that'll kill you.. and on 3 TB disks in a large raidz1 it can take days, even weeks to resilver onto a new disk, so your chance of that happening is not insignificant.
My personal recommendation to customers is to never use RAIDZ1 (RAID5 equivalent) at all with > 750 GB disks, ever, just to avoid a lot of potential unpleasantness. I've been OK with them breaking this rule because of other reasons (the system has a backup somewhere else, the data isn't that important, etc), but usually I do my best to push for RAIDZ2 as a minimum option with large disks.
Also, for a number of reasons, I usually recommend not going more than 8-12 disks in a raidz2 stripe or 11-15 disks in a raidz3 stripe. You should be on the low-end of those ranges with 3 TB disks, and could maybe be OK on the high-end of those ranges on 1 TB disks. That this will help keep you away from the idea that more disks will fail while a resilver is going on is only one of those reasons, but a big one.
If you're looking for some sane rules of thumb:
- Do not use raidz1 at all on > 750 GB disks.
- Do not use less than 3 or more than 7 disks on a raidz1.
- If thinking of using 3-disk raidz1 vdevs, seriously consider mirror vdevs instead.
- Do not use less than 6 or more than 12 disks on a raidz2.
- Do not use less than 7 or more than 15 disks on a raidz3.
- Always remember that unlike traditional RAID arrays where # of disks increase IOPS, in ZFS it is # of VDEVS, so going with shorter stripe vdevs improves pool IOPS potential.