当未指定 order by 时,SELECT TOP 如何工作?

2021-12-19 00:00:00 sql select tsql sql-server sql-order-by

功能,这意味着您的扫描将读取最近阅读的页面来自另一个并发扫描.

The msdn documentation says that when we write

SELECT TOP(N) ..... ORDER BY [COLUMN]

We get top(n) rows that are sorted by column (asc or desc depending on what we choose)

But if we don't specify any order by, msdn says random as Gail Erickson pointed out here. As he points out it should be unspecified rather then random. But as Thomas Lee points out there that

When TOP is used in conjunction with the ORDER BY clause, the result set is limited to the first N number of ordered rows; otherwise, it returns the first N number of rows ramdom

So, I ran this query on a table that doesn't have any indexes, first I ran this..

SELECT *
FROM
    sys.objects so
WHERE
    so.object_id NOT IN (SELECT si.object_id
                         FROM
                             sys.index_columns si)
    AND so.type_desc = N'USER_TABLE'

And then in one of those tables, (in fact I tried the query below in all of those tables returned by above query) and I always got the same rows.

SELECT TOP (2) *
FROM
    MstConfigSettings

This always returned the same 2 rows, and same is true for all other tables returned by query 1. Now the execution plans shows 3 steps..

As you can see there is no index look up, it's just a pure table scan, and

The Top shows actual no of rows to be 2, and so does the Table Scan; Which is not the case (there I many rows).

But when I run something like

SELECT TOP (2) *
FROM
    MstConfigSettings
ORDER BY
    DefaultItemId

The execution plan shows

and

So, when I don't apply ORDER BY the steps are different (there is no sort). But the question is how does this TOP works when there is no Sort and why and how does it always gives the same result?

解决方案

There is no guarantee which two rows you get. It will just be the first two retrieved from the table scan.

The TOP iterator in the execution plan will stop requesting rows once two have been returned.

Likely for a scan of a heap this will be the first two rows in allocation order but this is not guaranteed. For example SQL Server might use the advanced scanning feature which means that your scan will read pages recently read from another concurrent scan.

相关文章