Grassbell Deep discussion area bamboo
Registration Date: 2003 Jun From: Beijing, Xiamen Number of posts: 1018 |
Quote:
Originally released by Qinhuai night month
I feel that the select * from table is always scanned by block (it has not been verified by experiments, but it should be so in terms of efficiency ), the records inserted later may use the space left by the previous delete, so this is not necessarily in chronological order. In fact, relational databases emphasize physical independence. All operations only ensure the logic layer. If you want to extract rows in chronological order, you 'd better add a column to record the generation time of rows.
Select * from table return values: The rows in the same block are read in the order of row dictionary in the block header. Delete and insert again. Later, the row may use the deleted row in the row dictionary position. However, the order between blocks is probably determined by the order in which blocks is arranged in the segment. __________________ Donny's Email
Instead of your own, you can do more research, experiment, and write your own experiences.
|