相关疑难解决方法(0)

我构建了以下 SQL Server 查询，但它遇到了SQL Server 2005 中的反半连接缺陷，导致基数估计不准确（1 - 呃！）并永远运行。由于它是一个长期的生产 SQL Server，我不能轻易建议升级版本，因此我不能在这个特定查询上强制使用 traceflag 4199 提示。

我很难重构WHERE AND NOT IN (SELECT). 有人可以帮忙吗？我确保尝试使用基于集群密钥对的最佳连接。

SELECT TOP 5000 d.doc2_id
    ,d.direction_cd
    ,a.address_type_cd
    ,d.external_identification
    ,s.hash_value
    ,d.publishdate
    ,d.sender_address_id AS [D2 Sender_Address_id]
    ,a.address_id AS [A Address_ID]
    ,d.message_size
    ,d.subject
    ,emi.employee_id
FROM assentor.emcsdbuser.doc2 d(NOLOCK)
INNER JOIN assentor.emcsdbuser.employee_msg_index emi(NOLOCK)
    ON d.processdate = emi.processdate
    AND d.doc2_id = emi.doc2_id
INNER LOOP JOIN assentor.emcsdbuser.doc2_address a(NOLOCK)
    ON emi.doc2_id = a.doc2_id
    AND emi.address_type_cd = a.address_type_cd
    AND emi.address_id = a.address_id
INNER JOIN sis.dbo.sis s(NOLOCK) …