Excellent, thanks for that. Looks mostly like a legacy decision which wouldn't necessarily hold today.
Except for the part where you want integrity; AFAIK only xfs guarantees truly atomic snapshotting thanks to xfs_freeze.
Furthermore, it's also (by an extremely large margin) your best choice if you need a filesystem bigger than 16 TB.
Allegedly other filesystems have had a similar freeze added since then.
Actually it was added to the vfs layer IIRC, around 2.6.30.
Why wouldn't? is there a better alternative to freeze a filesystem and take a AWS snapshot from it?
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6... would indicate that the ability to freeze has been added to other filesystems since ec2-consistent-snapshot was first written. I've got no data on whether they're equivalently robust. If freezing was the only reason to go for xfs, then that decision would seem to be open to more alternatives now.