SqlBulkCopy 需要很长时间才能将数据表记录插入 SQL Server 表答案

【问题标题】：SqlBulkCopy takes long time to insert datatable records into SQL Server tableSqlBulkCopy 需要很长时间才能将数据表记录插入 SQL Server 表
【发布时间】：2015-09-11 11:11:17
【问题描述】：

我们正在尝试使用SqlBulkCopy.WriteToServer 方法插入 20000 条或更多记录，但插入表需要 140 秒。

有什么方法可以改善使用SqlBulkCopy 的时间，因为它广泛用于批量插入？

调用代码：

using (SqlBulkCopy bulkCopy = new SqlBulkCopy(dConn)) 
{
    bulkCopy.ColumnMappings.Add("SamplingFileId", "SamplingFileId");
    /*... 74 more columns ...*/
    bulkCopy.BulkCopyTimeout = 500; //Assign Destination Table Name
    bulkCopy.DestinationTableName = "cvr_ReviewSamplingLoans";
    bulkCopy.BatchSize = 5000; 
    bulkCopy.WriteToServer(dtLoansToUpload);
}

Sql表定义：

CREATE TABLE [dbo].[ReviewSamplingLoans](
[Id] [int] IDENTITY(1000,1) NOT NULL,
[SamplingFileId] [int] NOT NULL,
[ReviewId] [int] NOT NULL,
[LoanType] [nvarchar](10) NOT NULL,
[LoanNumber] [nvarchar](50) NOT NULL,
[BorrowerName] [nvarchar](150) NOT NULL,
[OriginalUPB] [money] NOT NULL,
[CurrentUPB] [money] NOT NULL,
[OriginalMonthlyPayment] [money] NULL,
[CurrentMonthlyP&I] [money] NULL,
[PaymentFrequency] [nvarchar](2) NULL,
[OriginalNoteRate] [float] NULL,
[CurrentInterestRate] [float] NULL,
[OriginationDate] [date] NULL,
[MaturityDate] [date] NULL,
[FirstPaymentDate] [date] NULL,
[InterestPaidThruDate] [date] NULL,
[NextPaymentDate] [date] NULL,
[DelinquencyStatus] [nvarchar](10) NULL,
[LatePaymentCount] [int] NULL,
[StatedTerm] [int] NULL,
[OriginalAmortizationTerm] [int] NULL,
[NoteType] [nvarchar](10) NULL,
[RateType] [nvarchar](10) NULL,
[LoanProductType] [nvarchar](10) NULL,
[BalloonDueDate] [date] NULL,
[BalloonPaymentAmount] [money] NULL,
[PropertyCity] [nvarchar](50) NULL,
[PropertyState] [nchar](2) NULL,
[PropertyZipcode] [nvarchar](10) NULL,
[PropertyType] [nvarchar](10) NULL,
[PropertyTypeDescription] [nvarchar](50) NULL,
[LienPosition] [tinyint] NULL,
[OriginalLTV] [float] NULL,
[CurrentLTV] [float] NULL,
[SeniorLienBalance] [money] NULL,
[OriginalCommitmentAmount] [money] NULL,
[CurrentCommitmentAmount] [money] NULL,
[OriginalCombinedLTV] [float] NULL,
[CurrentCombinedLTV] [float] NULL,
[OriginalPropertyValue] [money] NULL,
[CurrentPropertyValue] [money] NULL,
[AppraisalDate] [date] NULL,
[AppraisalType] [nvarchar](10) NULL,
[PMI] [nchar](1) NULL,
[LoanPurpose] [nchar](1) NULL,
[OccupancyType] [nchar](1) NULL,
[DocumentationType] [nvarchar](10) NULL,
[OriginalFICO] [smallint] NULL,
[CurrentFICO] [smallint] NULL,
[LastFICOUpdatedDate] [date] NULL,
[DebtToIncomeRatio] [float] NULL,
[InterestOnlyPeriod] [smallint] NULL,
[Modication] [nchar](1) NULL,
[Foreclosure] [nchar](1) NULL,
[ArmProductType] [nvarchar](10) NULL,
[ArmMargin] [float] NULL,
[ArmIndexType] [nvarchar](10) NULL,
[FirstRateAdjustmentDate] [date] NULL,
[NextRateAdjustmentDate] [date] NULL,
[LifeFloor] [float] NULL,
[LifeCeiling] [float] NULL,
[InitialRateCap] [float] NULL,
[PeriodicRateCap] [float] NULL,
[InternalRiskRating] [nvarchar](10) NULL,
[CurrentDebtServiceRatio] [float] NULL,
[CurrentNetOperIncome] [float] NULL,
[OriginalDebtServiceRatio] [float] NULL,
[OriginalNetOperIncome] [float] NULL,
[CurrentOccupancy] [float] NULL,
[PropertyTotalSquareFootage] [int] NULL,
[UnitCount] [smallint] NULL,
[PrepaymentPenalty] [nchar](1) NULL,
[PrepaymentPenaltyType] [int] NULL,
[PrepaymentPenaltyTerm] [smallint] NULL,
CONSTRAINT [PK_ReviewSamplingLoans] 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]

GO

ALTER TABLE [dbo].[ReviewSamplingLoans]  WITH NOCHECK ADD  CONSTRAINT [FK_ReviewSamplingLoans_ReviewCollateralTypes] FOREIGN KEY([LoanType])
REFERENCES [dbo].[ReviewSamplingLoans] ([Code])
GO

ALTER TABLE [dbo].[ReviewSamplingLoans] CHECK CONSTRAINT [FK_ReviewSamplingLoans_ReviewCollateralTypes]
GO

ALTER TABLE [dbo].[ReviewSamplingLoans]  WITH NOCHECK ADD  CONSTRAINT [FK_ReviewSamplingLoans_ReviewSamplingFiles] FOREIGN KEY([ReviewId])
REFERENCES [dbo].[Reviews] ([Id])
GO

ALTER TABLE [dbo].[ReviewSamplingLoans] CHECK CONSTRAINT [FK_ReviewSamplingLoans_ReviewSamplingFiles]
GO

关于大小，excel 文件包含 20000 行，这些行在验证时添加到数据表中，并使用 sqlbulkcopy 插入到表中。

【问题讨论】：

blogs.msdn.com/b/yingqin/archive/2014/02/28/…
没有任何具体信息很难做出回应。如何您使用 SqlBulkCopy？你在传递什么，一个DataTable，DataReader？记录有多大？网络快还是慢？你在使用交易吗？目标表是否有触发器？数据库的恢复模式是什么？
我们正在将数据表传递给SqlBulkCopy.WriteToServer，最小记录数为20000。网络相对较好，我们没有使用事务。
目标表不包含触发器。
好的，让我们从基础开始 1) 向我们展示您要插入的表的完整 SQL 定义，2) 您的网络的吞吐量是多少，3) 每个有多大（平均）记录？

标签： sql-server

【解决方案1】：

在选项上启用 Tablelock 可以更快地执行 SqlBulkCopy。

看看这里SqlBulkCopyOptions Enumeration

但请记住，一旦操作开始，在完成之前，任何人都无法使用该表。微软表示，这种技术提高了插入大量数据时的速度。

【讨论】：