所以我试图通过引用位于不同MYSQL服务器中的表来运行在远程MySQL服务器中创建表的代码。我尝试创建表的服务器有空间限制，并且被引用的表非常大，因此必须将它们保存在不同的远程服务器上。
我正在尝试找到一种方法来同时建立与两个数据库的持久连接（使用JDBC库），所以我不必保持缓冲少量的数据行......我想要能够直接引用数据。
E.g。 数据库A包含我引用的数据，数据库B是我创建新表的地方。 说我在数据库A中引用的表是1,000,000行。而不是打开与数据库A的连接，缓冲10,000行，关闭连接，打开与数据库B的连接，写入该数据库，删除我的缓冲区，然后重复......
我想与数据库A建立持久连接，因此每次写入数据库B都可以引用数据库A中的数据。
这可能吗？我已经尝试了几种方法（主要是通过创建新的连接对象，只有在连接断开时才刷新），而且我似乎无法使这个想法发挥作用。
有没有人使用JDBC做过类似的事情？如果是这样，如果你能指出我正确的方向，或者告诉我你是如何设法让它发挥作用的话，我将不胜感激。
答案 0 :(得分：1)
您可以在数据库A中创建数据，并通过复制将其复制到数据库B.
或者，听起来你正在实现某种队列。我曾经用Java构建了一个数据复制程序，它使用了Queue接口的内置实现。我有一个线程从数据库A中读取数据并填充队列，以及从队列中读取并写入数据库B的线程。我可以尝试挖掘我使用的类，如果有用的话吗？
编辑：
这是代码，稍微调整一下，用于发布。我没有包含配置类，但它应该让您了解如何使用队列类;
package test;
import java.io.File;
import java.sql.PreparedStatement;
import java.sql.ResultSet;
import java.sql.ResultSetMetaData;
import java.sql.SQLException;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.concurrent.ArrayBlockingQueue;
import java.util.concurrent.BlockingQueue;
/**
* This class implements a JDBC bridge between databases, allowing data to be
* copied from one place to another.
* <p>This implementation is threaded, as it uses a {@link BlockingQueue} to pass
* data between a producer and a consumer.
*/
public class DBBridge {
public static void main( String[] args ) {
Adaptor fromAdaptor = null;
Adaptor toAdaptor = null;
BridgeConfig config = null;
try {
/* BridgeConfig is essentially a wrapper around the Simple XML serialisation library.
* http://simple.sourceforge.net/
*/
config = BridgeConfig.loadConfig( new File( "db-bridge.xml" ) );
}
catch ( Exception e ) {
System.err.println( "Failed to read or parse db-bridge.xml: " + e.getLocalizedMessage() );
System.exit( 1 );
}
BlockingQueue<Object> b = new ArrayBlockingQueue<Object>( config.getQueueSize() );
try {
HashMap<String, DatabaseConfig> dbs = config.getDbs();
System.err.println( "Configured DBs" );
final String sourceName = config.getSource();
final String destinationName = config.getDestination();
if ( !dbs.containsKey( sourceName ) ) {
System.err.println( sourceName + " is not a configured database connection" );
System.exit( 1 );
}
if ( !dbs.containsKey( destinationName ) ) {
System.err.println( destinationName + " is not a configured database connection" );
System.exit( 1 );
}
DatabaseConfig sourceConfig = dbs.get( sourceName );
DatabaseConfig destinationConfig = dbs.get( destinationName );
try {
/*
* Both adaptors must be created before attempting a connection,
* as otherwise I've seen DriverManager pick the wrong driver!
*/
fromAdaptor = AdaptorFactory.buildAdaptor( sourceConfig, sourceConfig );
toAdaptor = AdaptorFactory.buildAdaptor( destinationConfig, destinationConfig );
System.err.println( "Connecting to " + sourceName );
fromAdaptor.connect();
System.err.println( "Connecting to " + destinationName );
toAdaptor.connect();
/* We'll send our updates to the destination explicitly */
toAdaptor.getConn().setAutoCommit( false );
}
catch ( SQLException e ) {
System.err.println();
System.err.println( "Failed to create and configure adaptors" );
e.printStackTrace();
System.exit( 1 );
}
catch ( ClassNotFoundException e ) {
System.err.println( "Failed to load JDBC driver due to error: " + e.getLocalizedMessage() );
System.exit( 1 );
}
DataProducer producer = null;
DataConsumer consumer = null;
try {
producer = new DataProducer( config, fromAdaptor, b );
consumer = new DataConsumer( config, toAdaptor, b );
}
catch ( SQLException e ) {
System.err.println();
System.err.println( "Failed to create and configure data producer or consumer" );
e.printStackTrace();
System.exit( 1 );
}
consumer.start();
producer.start();
}
catch ( Exception e ) {
e.printStackTrace();
}
}
public static class DataProducer extends DataLogger {
private BridgeConfig config;
private Adaptor adaptor;
private BlockingQueue<Object> queue;
public DataProducer(BridgeConfig c, Adaptor a, BlockingQueue<Object> bq) {
super( "Producer" );
this.config = c;
this.adaptor = a;
this.queue = bq;
}
@Override
public void run() {
/* The tables to copy are listed in BridgeConfig */
for ( Table table : this.config.getManifest() ) {
PreparedStatement stmt = null;
ResultSet rs = null;
try {
String sql = table.buildSourceSelect();
this.log( "executing: " + sql );
stmt = this.adaptor.getConn().prepareStatement( sql );
stmt.execute();
rs = stmt.getResultSet();
ResultSetMetaData meta = rs.getMetaData();
/* Notify consumer that a new table is to be processed */
this.queue.put( table );
this.queue.put( meta );
final int columnCount = meta.getColumnCount();
while ( rs.next() ) {
ArrayList<Object> a = new ArrayList<Object>( columnCount );
for ( int i = 0; i < columnCount; i++ ) {
a.add( rs.getObject( i + 1 ) );
}
this.queue.put( a );
}
}
catch ( InterruptedException ex ) {
ex.printStackTrace();
}
catch ( SQLException e ) {
// TODO Auto-generated catch block
e.printStackTrace();
}
try {
/* refresh the connection */
/* Can't remember why I can this line - maybe the other
* end kept closing the connection. */
this.adaptor.reconnect();
}
catch ( SQLException e ) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
try {
/* Use an object of a specific type to "poison" the queue
* and instruct the consumer to terminate. */
this.log( "putting finished object into queue" );
this.queue.put( new QueueFinished() );
this.adaptor.close();
}
catch ( InterruptedException e ) {
// TODO Auto-generated catch block
e.printStackTrace();
}
catch ( SQLException e ) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
}
/* Superclass for producer and consumer */
public static abstract class DataLogger extends Thread {
private String prefix;
public DataLogger(String p) {
this.prefix = p;
}
protected void log( String s ) {
System.err.printf( "%d %s: %s%n", System.currentTimeMillis(), this.prefix, s );
}
protected void log() {
System.err.println();
}
}
public static class DataConsumer extends DataLogger {
private BridgeConfig config;
private Adaptor adaptor;
private BlockingQueue<Object> queue;
private int currentRowNumber = 0;
private int currentBatchSize = 0;
private long tableStartTs = -1;
public DataConsumer(BridgeConfig c, Adaptor a, BlockingQueue<Object> bq) throws SQLException {
super( "Consumer" );
this.config = c;
this.adaptor = a;
this.queue = bq;
/* We'll send our updates to the destination explicitly */
this.adaptor.getConn().setAutoCommit( false );
}
public void printThroughput() {
double duration = ( System.currentTimeMillis() - this.tableStartTs ) / 1000.0;
long rowsPerSec = Math.round( this.currentRowNumber / duration );
this.log( String.format( "%d rows processed, %d rows/s", this.currentRowNumber, rowsPerSec ) );
}
@Override
public void run() {
this.log( "running" );
Table currentTable = null;
ResultSetMetaData meta = null;
int columnCount = -1;
PreparedStatement stmt = null;
while ( true ) {
try {
Object o = this.queue.take();
if ( o instanceof Table ) {
currentTable = (Table) o;
this.log( "processing" + currentTable );
if ( this.currentBatchSize > 0 ) {
/* Commit outstanding rows from previous table */
this.adaptor.getConn().commit();
this.printThroughput();
this.currentBatchSize = 0;
}
/* refresh the connection */
this.adaptor.reconnect();
this.adaptor.getConn().setAutoCommit( false );
/*
* Arguably, there's no need to flush the commit buffer
* after every table, but I like it because it feels
* tidy.
*/
this.currentBatchSize = 0;
this.currentRowNumber = 0;
if ( currentTable.isTruncate() ) {
this.log( "truncating " + currentTable );
stmt = this.adaptor.getConn().prepareStatement( "TRUNCATE TABLE " + currentTable );
stmt.execute();
}
this.tableStartTs = System.currentTimeMillis();
}
else if ( o instanceof ResultSetMetaData ) {
this.log( "received metadata for " + currentTable );
meta = (ResultSetMetaData) o;
columnCount = meta.getColumnCount();
String sql = currentTable.buildDestinationInsert( columnCount );
stmt = this.adaptor.getConn().prepareStatement( sql );
}
else if ( o instanceof ArrayList ) {
ArrayList<?> a = (ArrayList<?>) o;
/* One counter for ArrayList access, one for JDBC access */
for ( int i = 0, j = 1; i < columnCount; i++, j++ ) {
try {
stmt.setObject( j, a.get( i ), meta.getColumnType( j ) );
}
catch ( SQLException e ) {
/* Sometimes data in a shonky remote system
* is rejected by a more sane destination
* system. Translate this data into
* something that will fit. */
if ( e.getMessage().contains( "Only dates between" ) ) {
if ( meta.isNullable( j ) == ResultSetMetaData.columnNullable ) {
this.log( "Casting bad data to null: " + a.get( i ) );
stmt.setObject( j, null, meta.getColumnType( j ) );
}
else {
this.log( "Casting bad data to 0000-01-01: " + a.get( i ) );
stmt.setObject( j, new java.sql.Date( -64376208000L ), meta.getColumnType( j ) );
}
}
else {
throw e;
}
}
}
stmt.execute();
this.currentBatchSize++;
this.currentRowNumber++;
if ( this.currentBatchSize == this.config.getBatchSize() ) {
/*
* We've reached our non-committed limit. Send the
* requests to the destination server.
*/
this.adaptor.getConn().commit();
this.printThroughput();
this.currentBatchSize = 0;
}
}
else if ( o instanceof QueueFinished ) {
if ( this.currentBatchSize > 0 ) {
/* Commit outstanding rows from previous table */
this.adaptor.getConn().commit();
this.printThroughput();
this.log();
this.log( "completed" );
}
/* Exit while loop */
break;
}
else {
throw new RuntimeException( "Unexpected obeject in queue: " + o.getClass() );
}
}
catch ( InterruptedException e ) {
// TODO Auto-generated catch block
e.printStackTrace();
}
catch ( SQLException e ) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
try {
this.adaptor.close();
}
catch ( SQLException e ) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
}
public static final class QueueFinished {
/*
* This only exists as a completely type-safe value in "instanceof"
* expressions
*/
}
}
答案 1 :(得分：0)
我以前做过这个，我建议你做我做的，从DB A获取你需要的数据，然后把它写成一个或多个文件作为SQL'set'语句。当我这样做时，由于对DB B中加载文件大小的限制，我不得不拆分成大约10个文件
答案 2 :(得分：0)
在一个程序中，我写了一段时间回来工作，我有两个同时连接。如果没有提供代码，你就会想要
public void initialize() {
String dbUrl, dbUrl2, dbClass, dbClass2, user, user2, password, password2;
Connection con, con2;
Statement stmt, stmt2;
ResultSet rs, rs2;
try {
Class.forName(dbClass);
con = DriverManager.getConnection(dbUrl,user,password);
con2 = DriverManager.getConnection(dbUrl2,user2,password2);
stmt = con.createStatement();
} catch(ClassNotFoundException e) {
e.printStackTrace();
}
catch(SQLException e) {
e.printStackTrace();
}
}
然后，一旦你运行了两个连接，
rs = stmt.executeQuery("query");
rs2 = stmt2.executeQuery("second query");
我不知道如何专门解决您的问题，但这段代码可能会对您的系统造成一定的负担（假设您没有高端的个人/公司机器）并且可能需要一段时间。这应该足以让你至少开始，如果可能的话，我会发布更多内容，遗憾的是，模拟版本有点太复杂了。祝你好运！
答案 3 :(得分：0)
我认为通过两个独立的连接，一个读取连接和一个写连接，您将获得最佳服务，并使用一小部分缓冲区通过Java应用程序传递数据。
另一个更复杂但可能更优雅的解决方案是使用FEDERATED表。它使远程服务器上的表看起来是本地的。查询将传递到远程服务器并返回结果。您必须小心索引，否则它会非常慢，但它可能适用于您想要做的事情。
http://dev.mysql.com/doc/refman/5.5/en/federated-description.html