如何解决SQL Server中SQL身份危机

描述 (Description)

SQL Identity columns provide a convenient way to auto-number an ID column within a table without the need to manage the sequence. This convenience can save immense amounts of time, but also presents a single challenge: What happens when an identity column runs out of space within the data type chosen?

SQL Identity列提供了一种方便的方法来自动编号表中的ID列，而无需管理序列。这种便利可以节省大量时间，但是也带来了一个挑战：当标识列在所选数据类型内的空间不足时会发生什么？

In this article, we will answer this question, as well as explore a variety of ways to detect this problem before it arises and solve it without causing disruption to important tables.

在本文中，我们将回答这个问题，并探索各种方法以在出现问题之前对其进行检测并解决该问题，而不会对重要表造成干扰。

介绍 (Introduction)

SQL Identity columns are often used as a way to auto-number some data element when we have no need to assign any specific values to it. Either the values are arbitrary, the column is a surrogate key, or we wish to generate numbers for use in other processes downstream.

当不需要为数据元素分配任何特定值时，SQL Identity列通常用作自动编号某些数据元素的方法。值是任意的，列是代理键，或者我们希望生成数字以供下游的其他过程使用。

For most applications, identity columns are set-it-and-forget-it. For an integer, which can contain 2,147,483,647 positive values, it’s easy to assume that a number so large is out-of-reach for a fledgling table. As time goes on, though, and applications grow larger, two billion can quickly seem like a much smaller number. It’s important to understand our data usage, predict identity consumption over time, and proactively manage data type changes before an emergency arises.

对于大多数应用程序，标识列是设置后忘记的。对于一个可以包含2,147,483,647个正值的整数，很容易假设这么大的数字对于新创建的表是不可行的。但是，随着时间的流逝，应用程序越来越大，很快就会有20亿个数字变得很小。重要的是要了解我们的数据使用情况，预测一段时间内的身份消耗，并在紧急情况发生之前主动管理数据类型的更改。

SQL标识列用尽时会发生什么 (What happens when a SQL Identity Column is exhausted)

When a SQL identity column reaches its limit, all insert operations will fail. We can test this easily by creating a table with an identity column, reseeding it to its limit, and then trying to insert a new row. Here’s a table with only 2 columns, an identity that is set to a seed near its maximum value and a string:

当SQL标识列达到其限制时，所有插入操作将失败。我们可以通过创建一个带有标识列的表，将其重新设置到其极限，然后尝试插入新行来轻松测试。这是一个只有两列的表，一个身份设置为接近其最大值的种子和一个字符串：

CREATE TABLE dbo.Identity_Test
(   My_Identity INT NOT NULL IDENTITY(2147483646,1) CONSTRAINT PK_Identity_Test PRIMARY KEY CLUSTERED,My_Dinosaur VARCHAR(25) NOT NULL  );

With the table created, we can begin inserting some data:

创建表后，我们可以开始插入一些数据：


INSERT INTO dbo.Identity_Test(My_Dinosaur)
VALUES('Euoplocephalus');
INSERT INTO dbo.Identity_Test(My_Dinosaur)
VALUES('Triceratops');

After these inserts, we can view the data in our table:

这些插入之后，我们可以查看表中的数据：

Note that the identity value for our Triceratops is at the highest allowed by an integer data type. Now, let’s insert one more row:

请注意，三角龙的标识值是整数数据类型所允许的最高值。现在，让我们再插入一行：

INSERT INTO dbo.Identity_Test(My_Dinosaur)
VALUES('Micropachycephalosaurus');

The result of this SQL INSERT statement is the following error:

此SQL INSERT语句的结果是以下错误：

SQL Server provides no built-in warning. When we exceed the bounds of a data type, we receive the same error that would be returned if we tried to store a higher number in the INTEGER. Until an action is taken on our part to resolve the limit we have hit, inserts will continue to fail. All other operations on this table will execute normally, including DELETE, UPDATE, and SELECT.

SQL Server不提供内置警告。当我们超出数据类型的界限时，我们收到与试图将更大的数字存储在INTEGER中时将返回的相同错误。在我们采取行动解决我们遇到的限制之前，插入将继续失败。该表上的所有其他操作将正常执行，包括DELETE，UPDATE和SELECT。

The remainder of this article will deal with detecting a data type that is getting full with enough time so that we can take an action that does not need to be based on panic