Package
org.apache.commons.dbcp
Overview
Members
Books
SinceNot specified.
VersionNot specified.
AuthorNot specified.

Database Connection Pool API.

Overview in Dialog Form

Q: How do I use the DBCP package?

A: There are two primary ways to access the DBCP pool, as a Driver , or as a DataSource . You'll want to create an instance of PoolingDriver or PoolingDataSource . When using one of these interfaces, you can just use your JDBC objects the way you normally would. Closing a Connection will simply return it to its pool.

Q: But PoolingDriver and PoolingDataSource both expect an ObjectPool as an input. Where do I get one of those?

A: The ObjectPool interface is defined in the org.apache.commons.pool package (Commons-Pool). The org.apache.commons.pool.impl package has a couple of implementations, and you can always create your own.

Q: Ok, I've found an ObjectPool implementation that I think suits my connection pooling needs. But it wants a PoolableObjectFactory . What should I use for that?

A: The DBCP package provides a class for this purpose. It's called PoolableConnectionFactory . It implements the factory and lifecycle methods of PoolableObjectFactory for Connection s. But it doesn't create the actual database Connection s itself, if uses a ConnectionFactory for that. The PoolableConnectionFactory will take Connection s created by the ConnectionFactory and wrap them with classes that implement the pooling behaviour.

Several implementations of ConnectionFactory are provided--one that uses DriverManager to create connections ( DriverManagerConnectionFactory ), one that uses a Driver to create connections ( DriverConnectionFactory ), one that uses a DataSource to create connections ( DataSourceConnectionFactory ).

Q: I think I'm starting to get it, but can you walk me though it again?

A: Sure. Let's assume you want to create a DataSource that pools Connection s. Let's also assume that that those pooled Connection s should be obtained from the DriverManager . You'll want to create a PoolingDataSource .

The PoolingDataSource uses an underlying ObjectPool to create and store its Connection .

To create a ObjectPool , you'll need a PoolableObjectFactory that creates the actual Connection s. That's what PoolableConnectionFactory is for.

To create the PoolableConnectionFactory , you'll need at least two things:

  1. A ConnectionFactory from which the actual database Connection s will be obtained.
  2. An empty and factory-less ObjectPool in which the Connection s will be stored.
    When you pass an ObjectPool into the PoolableConnectionFactory , it will automatically register itself as the PoolableObjectFactory for that pool.
You can optionally provide a KeyedObjectPoolFactory that will be used to create KeyedObjectPool s for pooling PreparedStatement s for each Connection .

In code, that might look like this:

GenericObjectPool connectionPool = new GenericObjectPool(null);
DriverManagerConnectionFactory connectionFactory = new DriverConnectionFactory("jdbc:some:connect:string",null);
PoolableConnectionFactory poolableConnectionFactory = new PoolableConnectionFactory(connectionFactory,connectionPool,null,null,false,true);
PoolingDataSource dataSource = new PoolingDataSource(connectionPool);

To create a PoolingDriver , we do the same thing, except that instead of creating a DataSource on the last line, we create a PoolingDriver , and register the connectionPool with it. E.g.,:

GenericObjectPool connectionPool = new GenericObjectPool(null);
DriverManagerConnectionFactory connectionFactory = new DriverConnectionFactory("jdbc:some:connect:string",null);
PoolableConnectionFactory poolableConnectionFactory = new PoolableConnectionFactory(connectionFactory,connectionPool,null,null,false,true);
PoolingDriver driver = new PoolingDriver();
driver.registerPool("example",connectionPool);

Since the PoolingDriver registers itself with the DriverManager when it is created, now you can just go to the DriverManager to create your Connection s, like you normally would:

Connection conn = DriverManager.getConnection("jdbc:apache:commons:dbcp:example");

Q: Sounds complicated, is there an easier way?

A: If you're using the PoolingDriver , you don't need to do this configuration in code. Instead, you can provide a JOCL document that describes the connection pool, and let the PoolingDriver discover it at runtime.

Specifically, if the PoolingDriver is asked for a Connection from a pool that has not yet been registered, it will look for a named resource from which to read the pool's configuration, and create that pool.

For example, suppose you create a pool named "/eg" from a JOCL document. The "connect string" for this pool will be "jdbc:apache:commons:dbcp:/eg". To do this, you'll need a create a resource (just a file in your classpath) containing a JOCL description of the pool. Specifically, this JOCL document should define a PoolableConnectionFactory from which the pool will be obtained. For example:

<object class="org.apache.commons.dbcp.PoolableConnectionFactory" xmlns="http://apache.org/xml/xmlns/jakarta/commons/jocl">
   <!-- the first argument is the ConnectionFactory -->
   <object class="org.apache.commons.dbcp.DriverManagerConnectionFactory">
      <string value="jdbc:some:connect:string"/>
      <object class="java.util.Properties" null="true"/>
   </object>
   <!-- the next argument is the ObjectPool -->
   <object class="org.apache.commons.pool.impl.GenericObjectPool">
      <object class="org.apache.commons.pool.PoolableObjectFactory" null="true"/>
      <int value="10"/> <!-- max active -->
      <byte value="1"/> <!-- when exhausted action, 0 = fail, 1 = block, 2 = grow -->
      <long value="2000"/> <!-- max wait -->
      <int value="10"/> <!-- max idle -->
      <boolean value="false"/> <!-- test on borrow -->
      <boolean value="false"/> <!-- test on return -->
      <long value="10000"/> <!-- time between eviction runs -->
      <int value="5"/> <!-- number of connections to test per eviction run -->
      <long value="5000"/> <!-- min evictable idle time -->
      <boolean value="true"/> <!-- test while idle -->
   </object>
   <!-- the next argument is the KeyedObjectPoolFactory -->
   <object class="org.apache.commons.pool.StackKeyedObjectPoolFactory">
      <int value="5"/> <!-- max idle -->
   </object>
   <string value="SELECT COUNT(*) FROM DUAL"/> <;!-- validation query -->
   <boolean value="false"/> <!-- default read only -->
   <boolean value="true"/> <!-- default auto commit -->
</object>

Simply save that file somewhere in your classpath as eg.jocl, and the PoolingDriver will find it automatically. You need only register the PoolingDriver (for example, using the jdbc.drivers property), and use the the DriverManager to create your Connection s, like you normally would:

Connection conn = DriverManager.getConnection("jdbc:apache:commons:dbcp:/eg");

(Note that without the leading slash, the pool must be located at org/apache/commons/dbcp/PoolingDriver/eg.jocl within your classpath. See getResource(String) for details.)

Wiki javadoc Use textile entry format.

OK, so how about pooling preparedStatements? (is it what‘s actually needed to provide the same prepared statements available for all connections and hence threads? //if it makes sense .. )