HBase: HTablePool重构及优化

阅读更多

HBase Version: hbase-0.90.3-cdh3u1

org.apache.hadoop.hbase.client.HTablePool

用起来不是很方便. 所以重写了一些HTablePool, 对自己业务逻辑这块比较相关. 欢迎讨论.

主要是对源代码下面4点进行改进和设置:

1.  为不同的table建立的poolSize不一样, 目前HTablePool为所有的table建立的maxSize一致.

private final int maxSize;

2.  从HTablePool中getTable是只初始化了一个HTable, 而且在这个时候才初始化HTable的Queen.

    这个点不是很好, 也是我想修改源码的出发点.

  /**
   * Get a reference to the specified table from the pool.

   *
   * Create a new one if one is not available.
   * @param tableName table name
   * @return a reference to the specified table
   * @throws RuntimeException if there is a problem instantiating the HTable
   */
  public HTableInterface getTable(String tableName) {
    LinkedList queue = tables.get(tableName);
    if(queue == null) {
      queue = new LinkedList();
      tables.putIfAbsent(tableName, queue);
      return createHTable(tableName);
    }
    HTableInterface table;
    synchronized(queue) {
      table = queue.poll();
    }
    if(table == null) {
      return createHTable(tableName);
    }
    return table;
  }

 3.   应该有一个createTablePool的方法, 便于用户自己创建HTablePool.此方法可以与closeTablePool相互呼应.创建后面再关闭.

  public void closeTablePool(final String tableName)  {
    Queue queue = tables.get(tableName);
    synchronized (queue) {
      HTableInterface table = queue.poll();
      while (table != null) {
        this.tableFactory.releaseHTableInterface(table);
        table = queue.poll();
      }
    }
    HConnectionManager.deleteConnection(this.config, true);
  }

 4.  由于HTable的Put可以使用优化, 让多个Put一起提交flushCommits(). 循环pool的Htable,调用flushCommits().

 

贴上修改后的源码:

package org.apache.hadoop.hbase.client;

import java.util.LinkedList;
import java.util.Queue;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ConcurrentMap;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hbase.HBaseConfiguration;
import org.apache.hadoop.hbase.KeyValue;
import org.apache.hadoop.hbase.util.Bytes;
/**
 * A rewrite pool of HTable instances.

 *
 * Each HTablePool acts as a pool for all tables.  To use, instantiate an
 * HTablePool and use {@link #getTable(String)} to get an HTable from the pool.
 * Once you are done with it, return it to the pool with {@link #putTable(HTableInterface)}.
 * 
 * 
A pool can be created with a maxSize which defines the most HTable
 * references that will ever be retained for each table.  Otherwise the default
 * is {@link Integer#MAX_VALUE}.
 *
 * 
Pool will manage its own cluster to the cluster. See {@link HConnectionManager}.
 * 
 * @author greatwqs
 * @update 2012-08-25
 */
public class MyHTablePool {
	
	public final static int DEFAULT_POOL_SIZE   = 4;
	
	/**
	 * ConcurrentMap>
	 * String tableName
	 * LinkedList the HTable pool contains HTableInterface
	 * LinkedList you can create HTable pool different size you want.
	 */
    private final ConcurrentMap> tables 
                    = new ConcurrentHashMap>();
    /***
     * Configuration for hbase-site.xml
     */
	private final Configuration config;
	/**
	 * HTableInterfaceFactory that createHTableInterface and releaseHTableInterface
	 */
	private final HTableInterfaceFactory tableFactory;

	/**
	 * Default Constructor. 
	 */
	public MyHTablePool() {
		this(HBaseConfiguration.create());
	}

	/**
	 * Constructor to set maximum versions and use the specified configuration.
	 * @param config configuration
	 */
	public MyHTablePool(final Configuration config) {
		this(config, null);
	}

	public MyHTablePool(final Configuration config, final HTableInterfaceFactory tableFactory) {
		// Make a new configuration instance so I can safely cleanup when
		// done with the pool.
		this.config = config == null ? new Configuration() : new Configuration(
				config);
		this.tableFactory = tableFactory == null ? new HTableFactory()
				: tableFactory;
	}
	
	/**
	 * Create all the HTable instances , belonging to the given table.
	 * 

	 * Note: this is a 'create' of the given table pool.
	 * @param tableName 
	 * @param maxSize
	 * @param isAutoFlush
	 */
	public void createHTablePool(final String tableName, final int maxSize, boolean isAutoFlush) {
		LinkedList queue = tables.get(tableName);
		if (queue == null) {
			queue = new LinkedList();
			tables.putIfAbsent(tableName, queue);
		}
		synchronized (queue) {
			int addHTableSize = maxSize - queue.size();
			if(addHTableSize <= 0){
				return;
			}
			for(int i=0; i
	 * Note: this is a 'create' of the given table pool.
	 * @param tableNameArray
	 * @param maxSize
	 * @param isAutoFlush default false
	 * usage example:
	 * false: when {@link Put} use. use buffere put. call flushCommits after a time. 
	 *        you can design a thread(such as 3MS run a time)to loop all pool table, and call flushCommits.
	 *        the performance well.
	 * true: when {@link Scan} and {@link Delete} use. 
	 */
	public void createHTablePool(final String[] tableNameArray, final int maxSize, boolean isAutoFlush) {
		for(String tableName : tableNameArray){
			createHTablePool(tableName,maxSize,isAutoFlush);
		}
	}
	
	/**
	 * Create all the HTable instances , belonging to the given tables.
	 * 

	 * Note: this is a 'create' of the given table pool.
	 * @param tableName
	 * @param maxSize
	 */
	public void createHTablePool(final String[] tableNameArray, final int maxSize) {
		createHTablePool(tableNameArray,maxSize,false);
	}

	/**
	 * Get a reference to the specified table from the pool.

	 *
	 * @param tableName table name
	 * @return a reference to the specified table
	 * @throws RuntimeException if there is a problem instantiating the HTable
	 */
	public HTableInterface getHTable(String tableName) {
		LinkedList queue = tables.get(tableName);
		if (queue == null) {
			throw new RuntimeException("There is no pool for the HTable");
		}
		HTableInterface table;
		synchronized (queue) {
			table = queue.poll();
		}
		return table;
	}

	/**
	 * Get a reference to the specified table from the pool.

	 *
	 * Create a new one if one is not available.
	 * @param tableName table name
	 * @return a reference to the specified table
	 * @throws RuntimeException if there is a problem instantiating the HTable
	 */
	public HTableInterface getHTable(byte[] tableName) {
		return getHTable(Bytes.toString(tableName));
	}
	
	/**
	* Puts the specified HTable back into the pool.
    * 

    * 
    * If the HTable not belong to HTablePool before, do not use this method.
    * 
    * @param table table
    */
	public void putHTableBack(HTableInterface table) {
		LinkedList queue = tables.get(Bytes.toString(table.getTableName()));
		synchronized (queue) {
			queue.add(table);
		}
	}

	protected HTableInterface createHTable(String tableName) {
		return this.tableFactory.createHTableInterface(config, Bytes
				.toBytes(tableName));
	}

	/**
	 * Closes all the HTable instances , belonging to the given table, in the table pool.
	 * 

	 * Note: this is a 'shutdown' of the given table pool and different from
	 * {@link #putTable(HTableInterface)}, that is used to return the table
	 * instance to the pool for future re-use.
	 *
	 * @param tableName
	 */
	public void closeHTablePool(final String tableName) {
		Queue queue = tables.get(tableName);
		synchronized (queue) {
			HTableInterface table = queue.poll();
			while (table != null) {
				this.tableFactory.releaseHTableInterface(table);
				table = queue.poll();
			}
		}
		HConnectionManager.deleteConnection(this.config, true);
	}

	/**
	 * See {@link #closeTablePool(String)}.
	 *
	 * @param tableName
	 */
	public void closeHTablePool(final byte[] tableName) {
		closeHTablePool(Bytes.toString(tableName));
	}
	
	/**
	 * See {@link #closeTablePool(String)}.
	 *
	 * @param tableName
	 */
	public void closeHTablePool() {
		for(String tabName:tables.keySet()){
			closeHTablePool(tabName);
		}
	}

	/**
	 * getCurrentPoolSize
	 * @param tableName
	 * @return
	 */
	public int getCurrentPoolSize(String tableName) {
		Queue queue = tables.get(tableName);
		synchronized (queue) {
			return queue.size();
		}
	}
	
} 

 

org.apache.hadoop.hbase.client.MyHTablePoolTest 测试实例

package org.apache.hadoop.hbase.client;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hbase.KeyValue;

/**
 * @author greatwqs
 * @update 2012-08-25
 */
public class MyHTablePoolTest {
	
	/**
	 * test method
	 * @param args
	 * @throws Exception
	 */
	public static void main(String[] args) throws Exception{
		// 1. my config file 
		String configFile = "conf/hbase-site.xml";
		Configuration config = new Configuration();
		config.addResource(new Path(configFile));
		// 2. init HTablePool
		MyHTablePool myPool = new MyHTablePool(config);
		// 3. create HTablePool for a table
		myPool.createHTablePool("DCP_DataCenter_Base", MyHTablePool.DEFAULT_POOL_SIZE, false);
		// 4. get already exist HTable from HTablePool
		HTable table = (HTable) myPool.getHTable("DCP_DataCenter_Base");
		if(table != null){
			System.out.println("get HTable from HTablePool Success!");
		}
		// 5. get all data from HTable, and print to console.
		Scan scan = new Scan();
		ResultScanner rs = table.getScanner(scan);
		try {
			for (Result result : rs) {
				KeyValue[] kv = result.raw();
				byte[] key = kv[0].getRow();
				System.out.println("RowKey: " + new String(key));
				for (int i = 0; i < kv.length; i++) {
					System.out.println("ColumnFamily: "	+ new String(kv[i].getFamily()));
					System.out.println("Qualifier: "+ new String(kv[i].getQualifier()));
					System.out.println("Timestamp: "+ String.valueOf(kv[i].getTimestamp()));
					System.out.println("Value: " + new String(kv[i].getValue()));
				}
				System.out.println();
			}
		} catch (Exception e) {
			e.printStackTrace();
		} finally {
			rs.close();
		}
		// 6. after use HTable end, then put the HTable back to HTablePool.
		myPool.putHTableBack(table);
		// 7. close HTablePool
		myPool.closeHTablePool();
	}
}