SQLPlUS Copy命令學習筆記_20110806

gdutllf2006發表於2011-08-07

Oracle 10g開始在SQLPLUS中引入了Copy命令

SQLPLUS Copy命令可以在不同的資料庫之間或不同的使用者之間複製表資料,且錶帶有Long欄位不能使用CTAS時相當有用;

示例:
sqlplus /nolog
set long 1000       
set copycommit 10   
set arraysize 5000  
copy from to create new_t using select * from t;
copy from to insert new_t using select * from t;
copy from to append new_t using select * from t;
copy from to replace new_t using select * from t;


Copy Command
The SQL*Plus COPY command can copy data between two databases via SQL*Net .  The preferred method of doing this is to use SQL*Plus on the host where the database resides.  If performing the copy command from a client SQL*Net connection, the data is transferred through the client machine.

The copy command copies data from one Oracle instance to another.   The data is simply copied directly from a source to a target.  The format of the copy command is:

COPY FROM database TO database action -  destination_table (column_name, column_name...) USING query

The action can include:

1) create – If the destination table already exists, copy will report an error, otherwise the table is created and the data is copied. 

2) replace – If the destination table exists, copy will drop and recreate the table with the newly copied data. Otherwise, it will create the table and populate it with the data. 

3) insert – If the destination table exists, copy inserts the new rows into the table. Otherwise, copy reports an error and aborts.

4) append– Inserts the data into the table if it exists, otherwise it will create the table and then insert the data. 


 
#可以設定的引數
Once the command above is executed, the copy utility displays the values of three parameters, each of which can be set with the SQL*Plus set command. 
The arraysize specifies the number of rows that SQL*Plus will retrieve from the database at one time. 
The copycommit parameter specifies how often a commit is performed and is related to the number of trips – one trip is the number of rows defined in arraysize. 
The long parameter displays the maximum number of characters copied for each column with a LONG datatype.


#設定不同的列名
The command above did not specify column names for the new table (new_t).  As a result, the new table will have the same column names as the table being copied.  If different column names are required, they can be specified after the table name:

SQL>copy from to create new_t using(col1,col2,col3,....) select * from t;

However, if one column name is specified, they all must be specified. 

A DBA could perform. this same function with a database link from one database pointing to another.  The appeal of the copy command is that it only requires SQL*Net service names and proper privileges to get the job done.  For those environments that restrict the usage of database links, the copy utility can be leveraged.  In addition, the copy command provides many options, as defined by the actions create, replace, insert and append. 

If the copy command is executed from a client PC to copy data from remote database DB0 to remote database DB1, the data will be copied from DB0 to the client PC and then to DB1.  For this reason, it is best to use SQL*Plus from either remote host and not require the data to travel through a client machine in order to reach its final destination.

參考文件<>

 

來自 “ ITPUB部落格 ” ,連結:http://blog.itpub.net/10248702/viewspace-704339/,如需轉載,請註明出處,否則將追究法律責任。

相關文章