The suspected configuration is a 3 node system using S3 storage. We need to determine what topologies such a slowdown occurs:
1) multi-node local storage
2) multi-node S3
3) single node local storage
4) single node S3
The scientific method shall be used to determine which of the above manifests a DDL slowdown between 1.2 and 5.5. Since Storage Manager isn't available for 1.2, we sahll collect empirical data from all available configurations for the versions. We may need to throw 5.4 in the mix to get some more data points.

The purpose of this exercise is narrow the scope of the next step, which is to determine in what code the slowdown occurs.

An interesting thing arose during my discussions today. It appears that under 3 node AWS with S3 storage, a CREATE TABLE statement was consistent in taking exactly 6 seconds. This consistency I find suspicious. It's as if a timeout was being waited on. I don't know if this helps, but it wouldn't hurt to keep in the back of your mind as the investigation proceeds.

This seems to be a symptom of a shared metadata folder like /var/lib/columnstore/storagemanager via EFS.