插入时的索引性能

Question

我有一个有 4 列的表，我在这个表上有 3 个索引：

   CREATE TABLE [dbo].CustomerInfo(
        ID [int] IDENTITY(1,1) NOT NULL,
        UserHashID [varchar](20) NOT NULL,
        ShippingID [int] NOT NULL,
        Received [bit] NOT NULL,
     CONSTRAINT [PK_ID] PRIMARY KEY CLUSTERED 
    (
        [ID] ASC
    )WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON) ON [PRIMARY]
    ) ON [PRIMARY]

CREATE NONCLUSTERED INDEX [IDX_CustomerInfo_ShippingID] ON [dbo].[CustomerInfo]  (  [ShippingID] ASC )WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, SORT_IN_TEMPDB = OFF, IGNORE_DUP_KEY = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON) ON [PRIMARY]

CREATE NONCLUSTERED INDEX [IDX_CustomerInfo_UserHashID] ON [dbo].[CustomerInfo]  (  [UserHashID] ASC )WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, SORT_IN_TEMPDB = OFF, IGNORE_DUP_KEY = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON) ON [PRIMARY]

CREATE NONCLUSTERED INDEX [IDX_CustomerInfo_UserHashIDShippingID] ON [dbo].[CustomerInfo]  (    [UserHashID] ASC,   [ShippingID] ASC )WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, SORT_IN_TEMPDB = OFF, IGNORE_DUP_KEY = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON) ON [PRIMARY]

我插入了大约 4-5 百万条记录，这个过程大约需要 45 分钟。我意识到如果我删除索引，插入速度会更快（2-3 分钟）。

想知道删除索引、插入并在插入完成后重建索引是否有任何副作用。与启用索引相比，整个过程可能需要 5 分钟，而如果启用索引则需要 45 分钟。

Answer 1

不，删除索引、执行插入和在插入完成后重新构建索引没有副作用（假设在执行插入时没有其他东西需要访问表）。

这是很常见的模式。

[话虽如此，我对具有 4 列和 3 个索引的表的时差感到惊讶。你能发布你的架构和索引定义吗]

正如@PJ8912 所指出的，事务日志记录可能会有所不同，具体取决于您备份事务日志的频率。

更新：不相关，但是这个索引

CREATE NONCLUSTERED INDEX [IDX_CustomerInfo_UserHashID] 
    ON [dbo].[CustomerInfo]  ([UserHashID] ASC)

是多余的，因为它被这个索引覆盖：

CREATE NONCLUSTERED INDEX [IDX_CustomerInfo_UserHashIDShippingID] 
    ON [dbo].[CustomerInfo]  ([UserHashID] ASC, [ShippingID] ASC)

Answer 2

根据事务日志记录的级别，TLogs 可能会定期重新创建索引而填满。如果您截断索引以消除它们，则不会记录该操作。

创建新索引后，执行计划的统计信息可能不是最新的。您可能希望使用 FULL SCAN 模式更新统计信息。