Question

使用带有标识和日期 ON分区的PRIMARY KEY CLUSTERED创建了两个表，因为任务会拆分并截断早于30天的分区。仅保留最近的记录。

我有一个特殊字段来关联表格，但即使使用索引，加入也会非常缓慢。你能建议如何优化吗？

接下来是表和Join语句：

CREATE TABLE [dbo].[Redeem](
    [Id] [int] IDENTITY(1,1) NOT NULL,
    [Ticket] [nvarchar](64) NOT NULL,
    [CorrelationTicket] [nvarchar](64) NOT NULL,
    [CreatedUTC] [datetime] NOT NULL,
    [CreatedDate] [date] NOT NULL,
    [Redeem fields here...]
 CONSTRAINT [PK_Redeem] PRIMARY KEY CLUSTERED 
(
    [CreatedDate] ASC,
    [Id] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON)
)
ON myPS([CreatedDate]);

CREATE NONCLUSTERED INDEX [IX_Redeem_CorrelationTicket]
ON [dbo].[Redeem]([CreatedDate] ASC, [CorrelationTicket] ASC)
ON [myPS] ([CreatedDate]);

CREATE TABLE [dbo].[Validate](
    [Id] [int] IDENTITY(1,1) NOT NULL,
    [Ticket] [nvarchar](64) NOT NULL,
    [CorrelationTicket] [nvarchar](64) NOT NULL,
    [CreatedUTC] [datetime] NOT NULL,
    [CreatedDate] [date] NOT NULL,
    [Validate fields here...]
 CONSTRAINT [PK_Validate] PRIMARY KEY CLUSTERED 
(
    [CreatedDate] ASC,
    [Id] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON)
)
ON myPS([CreatedDate]);

CREATE NONCLUSTERED INDEX [IX_Validate_CorrelationTicket]
ON [dbo].[Validate]([CreatedDate] ASC, [CorrelationTicket] ASC)
ON [myPS] ([CreatedDate]);

这是加入：

SELECT top 100
    v.*,
    r.*
from 
    Validate v
LEFT OUTER join Redeem r
    on v.CorrelationTicket = r.CorrelationTicket
ORDER BY v.CreatedDate DESC

Answer 1

谢谢Krintner！

问题是您建议的ORDER BY。它正在整理整个结果集（不确定原因）。我确实需要排序，但将索引更改为DESC就可以了。

我也遵循建议并仅在INDEX中使用CorrelationTicket。

CONSTRAINT [PK_Redeem] PRIMARY KEY CLUSTERED 
(
    [CreatedDate] DESC,
    [Id] DESC
)

具有主键聚簇索引的表之间的关系和连接

1 个答案: