一、创建一个用来测试的数据库和表
1 USE [Test] 2 GO 3 /****** Object: Table [dbo].[student] Script Date: 2019/4/11 15:38:59 ******/ 4 SET ANSI_NULLS ON 5 GO 6 SET QUOTED_IDENTIFIER ON 7 GO 8 CREATE TABLE [dbo].[student]( 9 [id] [int] IDENTITY(1,1) NOT NULL, 10 [name] [nvarchar](50) NULL, 11 [age] [int] NULL, 12 [createDate] [datetime] NULL, 13 CONSTRAINT [PK_student] PRIMARY KEY CLUSTERED 14 ( 15 [id] ASC 16 )WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY] 17 ) ON [PRIMARY] 18 19 GO
方式1：一条一条插入，性能最差
1 INSERT INTO student(name,age,createDate) VALUES('苏拉自摸1',1,'2018-4-11'); 2 INSERT INTO student(name,age,createDate) VALUES('苏拉自摸2',2,'2018-4-11');
1 public void InsertOne() 2 { 3 Console.WriteLine("采用一条一条插入的方式实现"); 4 Stopwatch sw = new Stopwatch(); 5 6 using (SqlConnection conn = new SqlConnection(Conn)) //using中会自动Open和Close 连接。 7 { 8 string sql = "INSERT INTO student(name,age,createDate) VALUES(@a,@p,@d)"; 9 conn.Open(); 10 for (int i = 0; i < 1000000; i++) 11 { 12 using (SqlCommand cmd = new SqlCommand(sql, conn)) 13 { 14 cmd.Parameters.AddWithValue("@a", "花拉子模" + i); 15 cmd.Parameters.AddWithValue("@p", i); 16 cmd.Parameters.AddWithValue("@d", DateTime.Now); 17 sw.Start(); 18 cmd.ExecuteNonQuery(); 19 Console.WriteLine(string.Format("插入{0}条记录，已耗时{1}毫秒",i, sw.ElapsedMilliseconds)); 20 } 21 if (i == 1000) 22 { 23 sw.Stop(); 24 break; 25 } 26 } 27 } 28 Console.WriteLine(string.Format("插入{0}条记录，每{4}条的插入时间是{1}毫秒,预估总得插入时间是{2}毫秒，{3}分钟", 29 1000000, sw.ElapsedMilliseconds, ((sw.ElapsedMilliseconds / 1000) * 1000000), GetMinute((sw.ElapsedMilliseconds / 1000 * 1000000)), 1000)); 30 }
运行结果：
100w条记录，插完要俩小时了
方式2：使用sqlBulk
语法如下：
1 BULK INSERT [ [ 'database_name'.][ 'owner' ].]{ 'table_name' FROM 'data_file' } 2 WITH ( 3 [ BATCHSIZE [ = batch_size ] ], 4 [ CHECK_CONSTRAINTS ], 5 [ CODEPAGE [ = 'ACP' | 'OEM' | 'RAW' | 'code_page' ] ], 6 [ DATAFILETYPE [ = 'char' | 'native'| 'widechar' | 'widenative' ] ], 7 [ FIELDTERMINATOR [ = 'field_terminator' ] ], 8 [ FIRSTROW [ = first_row ] ], 9 [ FIRE_TRIGGERS ], 10 [ FORMATFILE = 'format_file_path' ], 11 [ KEEPIDENTITY ], 12 [ KEEPNULLS ], 13 [ KILOBYTES_PER_BATCH [ = kilobytes_per_batch ] ], 14 [ LASTROW [ = last_row ] ], 15 [ MAXERRORS [ = max_errors ] ], 16 [ ORDER ( { column [ ASC | DESC ] } [ ,...n ] ) ], 17 [ ROWS_PER_BATCH [ = rows_per_batch ] ], 18 [ ROWTERMINATOR [ = 'row_terminator' ] ], 19 [ TABLOCK ], 20 )
相关参数说明：
1 BULK INSERT 2 [ database_name . [ schema_name ] . | schema_name . ] [ table_name | view_name ] 3 FROM 'data_file' 4 [ WITH 5 ( 6 [ [ , ] BATCHSIZE = batch_size ] --BATCHSIZE指令来设置在单个事务中可以插入到表中的记录的数量 7 [ [ , ] CHECK_CONSTRAINTS ] --指定在大容量导入操作期间，必须检查所有对目标表或视图的约束。若没有 CHECK_CONSTRAINTS 选项，则所有 CHECK 和 FOREIGN KEY 约束都将被忽略，并且在此操作之后表的约束将标记为不可信。 8 [ [ , ] CODEPAGE = { 'ACP' | 'OEM' | 'RAW' | 'code_page' } ] --指定该数据文件中数据的代码页 9 [ [ , ] DATAFILETYPE = 10 { 'char' | 'native'| 'widechar' | 'widenative' } ] --指定 BULK INSERT 使用指定的数据文件类型值执行导入操作。 11 [ [ , ] FIELDTERMINATOR = 'field_terminator' ] --标识分隔内容的符号 12 [ [ , ] FIRSTROW = first_row ] --指定要加载的第一行的行号。默认值是指定数据文件中的第一行 13 [ [ , ] FIRE_TRIGGERS ] --是否启动触发器 14 [ [ , ] FORMATFILE = 'format_file_path' ] 15 [ [ , ] KEEPIDENTITY ] --指定导入数据文件中的标识值用于标识列 16 [ [ , ] KEEPNULLS ] --指定在大容量导入操作期间空列应保留一个空值，而不插入用于列的任何默认值 17 [ [ , ] KILOBYTES_PER_BATCH = kilobytes_per_batch ] 18 [ [ , ] LASTROW = last_row ] --指定要加载的最后一行的行号 19 [ [ , ] MAXERRORS = max_errors ] --指定允许在数据中出现的最多语法错误数，超过该数量后将取消大容量导入操作。 20 [ [ , ] ORDER ( { column [ ASC | DESC ] } [ ,...n ] ) ] --指定数据文件中的数据如何排序 21 [ [ , ] ROWS_PER_BATCH = rows_per_batch ] 22 [ [ , ] ROWTERMINATOR = 'row_terminator' ] --标识分隔行的符号 23 [ [ , ] TABLOCK ] --指定为大容量导入操作持续时间获取一个表级锁 24 [ [ , ] ERRORFILE = 'file_name' ] --指定用于收集格式有误且不能转换为 OLE DB 行集的行的文件。 25 )]
代码：
1 public void InsertTwo() 2 { 3 Console.WriteLine("使用Bulk插入的实现方式"); 4 Stopwatch sw = new Stopwatch(); 5 DataTable dt = GetTableSchema(); 6 7 using (SqlConnection conn = new SqlConnection(Conn)) 8 { 9 SqlBulkCopy bulkCopy = new SqlBulkCopy(conn); 10 bulkCopy.DestinationTableName = "student"; 11 bulkCopy.BatchSize = dt.Rows.Count; 12 conn.Open(); 13 sw.Start(); 14 15 for (int i = 0; i < 1000000; i++) 16 { 17 DataRow dr = dt.NewRow(); 18 dr["name"] = "蔑儿乞人" + i; 19 dr["age"] = i; 20 dr["createDate"] = DateTime.Now.ToString("yyyy-MM-dd HH:mm:ss"); 21 dt.Rows.Add(dr); 22 } 23 if (dt != null && dt.Rows.Count != 0) 24 { 25 bulkCopy.WriteToServer(dt); 26 sw.Stop(); 27 } 28 Console.WriteLine(string.Format("插入{0}条记录共花费{1}毫秒，{2}分钟", 1000000, sw.ElapsedMilliseconds, GetMinute(sw.ElapsedMilliseconds))); 29 } 30 } 31 public DataTable GetTableSchema() 32 { 33 DataTable dt = new DataTable(); 34 dt.Columns.AddRange( 35 new DataColumn[] { 36 new DataColumn("name",typeof(string)), 37 new DataColumn("age",typeof(int)), 38 new DataColumn("createDate",typeof(DateTime ))}); 39 return dt; 40 }
方式3：insert into
INSERT INTO student(name,age,createDate) SELECT '托多1'1,'2019-4-11', UNION ALL SELECT '托多2'2,'2019-4-11', UNION ALL ......
四、拼接sql
1 INSERT INTO student(name,age,createDate)VALUES
2 ('牛栏1段',160,'2019-4-11')
3 ,('牛栏2段',260,'2019-4-11')
4 ......