If it is a single file, the S3-write is going to be a single commit and a single S3 block. S3 isn't a filesystem
Aye, there's the rub.
If you need a network-backed filesystem, use a network backed filesystem (e.g., EFS). Trying to use S3-backed storage that way is going to be painfully slow.
Corollary: volumes restored from EBS snapshots can be awfully slow, because EBS doesn't grab the entire volume from S3 when it "restores" it - it only grabs blocks on demand.
danno
