Dictionary changes and update sets Usually, using update sets prevent you from applying dictionary changes that result in data loss. Blocked dictionary changes include: Removing tables Changing a column data type Update sets do not track the removal of tables from the system dictionary. Instead, customers must manually remove tables from the target instance. While update sets track data type changes, the target instance skips any change that results in data loss and instead adds a log message about the action. Customers can use the log to manually make data type changes on the target instance. Note: Update set previews do not check for type mismatch problems since the target instance skips changes resulting in data loss. Also, using update sets to delete a column from a table can cause data loss in certain circumstances. If there is data in the column on the target instance, that data is deleted, as well as the column itself, when the update set is committed.