The sampling is random across data pages, and is taken from the table or the smallest nonclustered index on the columns needed by the statistics.

• 采样是在各个数据页上随机进行的,取自表或统计信息所需列的最小非聚集索引。