Very uneven disk size in btrfs RAID-1

Question

I want to grow my 6x 3TB RAID-1, and want to use the chance to start using bigger disk sizes. I know that I can use uneven disk sizes in btrfs arrays, and as long as the biggest disk isn't larger than 50% of the array, I should also be able to use the entire space (calculator).

But besides usable disk space, what other consequences will I have to face? E.g. the largest disk will be hit with much more write load than the other disks, no? What about read load? Is this a thing I should keep in mind? Could this impact the reliability of the array?

Besides maximum possible disk size difference, is there a recommended maximum disk size difference for these scenarios?

Answer 1

But besides usable disk space, what other consequences will I have to face? E.g. the largest disk will be hit with much more write load than the other disks, no? What about read load? Is this a thing I should keep in mind? Could this impact the reliability of the array?

If the 10TB disk holds copies of data corresponding to ~three 3TB disks, it'll likely also have ~3× the load as well. (This is purely a guesstimate though.)I don't think reliability will be a problem at all – but performance might be, as the 10TB disk might become the bottleneck.

(If the same data is stored on two disks, then I believe each disk will have 1⁄2 probability of being read from. Currently, Btrfs has a fairly simplistic read balancing mechanism based on the process ID; as I understand it, the same process will always be directed to the same disk, approximately.)