SQL 语句 - 如何通过索引提高速度

Question

我有一个脚本，它必须查看超过 250 万条记录，以查找某个成员是否有未读电子邮件。我想知道可以做些什么来提高它的速度。目前，运行脚本最多可能需要 8 秒:

SELECT TOP(1) MemberMailID
FROM MemberMail
WHERE ToReadFlag = 0
AND ToMemberID = 102
AND ToDeletedFlag = 0
AND FromDeletedFlag = 0
AND OnHold = 0
AND ToArchivedFlag = 0

如何使用索引使其更快？

Answer 1

这个索引可能会有帮助，但请记住天下没有免费的午餐(必须维护索引，因此这会影响您的插入/更新/删除工作量):

CREATE NONCLUSTERED INDEX unread_emails
  ON dbo.MemberMail(ToMemberID)
  INCLUDE (MemberMailID)
  WHERE ToReadFlag = 0
  AND ToDeletedFlag = 0
  AND FromDeletedFlag = 0
  AND OnHold = 0
  AND ToArchivedFlag = 0;

现在您的查询可以说:

SELECT TOP (1) MemberMailID
  FROM dbo.MemberMail -- dbo prefix
    WITH (INDEX (unread_emails)) -- in case you need to force, though you should not
WHERE ToMemberID = 102
AND ToReadFlag = 0
AND ToDeletedFlag = 0
AND FromDeletedFlag = 0
AND OnHold = 0
AND ToArchivedFlag = 0
ORDER BY ToMemberID; -- ORDER BY is important!

如果您根据查询更改其中一些标志的值，您可以尝试将这些列添加到索引的键而不是过滤器，例如假设有时您检查 OnHold = 0，有时检查 OnHold = 1:

CREATE NONCLUSTERED INDEX unread_emails
  ON dbo.MemberMail(ToMemberID, OnHold)
  INCLUDE (MemberMailID)
  WHERE ToReadFlag = 0
  AND ToDeletedFlag = 0
  AND FromDeletedFlag = 0
  AND ToArchivedFlag = 0;

您可能还想尝试在 key 中使用 MemberMailID 而不是 INCLUDE。例如:

CREATE NONCLUSTERED INDEX unread_emails
  ON dbo.MemberMail(ToMemberID, MemberMailID)
  WHERE ToReadFlag = 0
  AND ToDeletedFlag = 0
  AND FromDeletedFlag = 0
  AND OnHold = 0
  AND ToArchivedFlag = 0;

这些差异可能对您的数据和使用模式无关紧要，但您将能够比我们猜测的更容易地测试差异。