简体繁体 English

将大型Nvarchar（max）表从一个数据库复制到另一个数据库

[英]copy large Nvarchar(max) table from one db to another

原文 2017-07-26 04:45:32 0 1 sql/ ssis/ ssis-2008

I have a table with more then 60 fields of Nvarchar(max) and i need to copy the table to another db with ssis. 我有一个具有Nvarchar（max）的60多个字段的表，我需要将该表复制到具有sis的另一个db。

i disabled all the index's and it's helped a bit, but still it's take a lot of time to copy all the table (more then 1M rows) - it's take an hour for this moment. 我禁用了所有索引，这有所帮助，但是复制所有表（超过1M行）仍然需要花费很多时间-这需要一个小时。

Do anyone have an idea to make it run faster? 有没有人想让它运行更快？

1 个解决方案

In my experience, the fastest way to load large data is to use bulk insert. 以我的经验，加载大数据的最快方法是使用批量插入。 This is usually done from a flat file, so you have to export the data to a file first, before performing the bulk insert. 通常，这是从平面文件完成的，因此在执行批量插入之前，您必须先将数据导出到文件中。

If you don't like the flat file approach, the alternative is to write a CLR routine. 如果您不喜欢平面文件方法，则可以选择编写CLR例程。 You can use a SqlDataReader to read from table storing the data in a DataTable and then using SqlBulkCopy to do the bulk insert. 您可以使用SqlDataReader从将数据存储在DataTable中的表中读取，然后使用SqlBulkCopy进行批量插入。 The advantage of this approach is that you can use a buffer system to limit the memory required. 这种方法的优点是可以使用缓冲系统来限制所需的内存。 The SqlDataReader is memory efficient, it reads lines as required and then drops them from memory, the DataTable is not. SqlDataReader的内存效率很高，它会根据需要读取行，然后将其从内存中删除，而DataTable则不是。 So you set yourself a limit, say 10,000 rows, and whenever the DataTable hits this limit you perform SqlBulkCopy and delete the existing rows in the DataTable and carry on with the next batch. 因此，您为自己设置了一个限制，例如10,000行，并且只要DataTable达到此限制，就执行SqlBulkCopy并删除DataTable中的现有行，然后继续进行下一批处理。