构建更有效的WEB应用

　　Dispatcher生存在一个独立的线程中并且是所有并发Job的容器。每一批处理在两种条件下会执行，不考虑先后:

　　1、 自上一次这种条件下执行后mIntervalTime毫秒并且至少有一个Job在批次中。

　　2、 批次的大小达到mJobBatchMaxSize

　　如果mJobBatchMaxSize值设为1，TrainServlet就退化为ClassicServlet。这是一个动态批处理的重要标识，这会在稍后解释。

　　方案的最后部分是Worker类:

//Worker.java
package train;

import java.util.*;
import java.sql.*;

public class Worker
    implements Runnable {
  List mJobs;
  Map mJobMap; //Helper member for mapping the jobs
  protected Worker(List jobs) {
    mJobs = jobs;
    mJobMap = new HashMap();
  }

  private void Process() {
    boolean isError = false;
    StringBuffer sqlBuff = new StringBuffer(
        "Select id,name ,descr,views from trentry where id in ");
    StringBuffer whereClause = CreateWhereClause(); //
    sqlBuff.append(whereClause);
    /* Now SQL statement is fully formed and looks like:

     Select id,name ,descr,views from trentry where id in (3343,22222,5555).
         This will allow us to fetch several user requests in one shot */
    Connection connection = null;
    Statement stmt = null;
    try {
      connection = Util.getDBConnection();
      stmt = connection.createStatement();
      ResultSet rs = stmt.executeQuery(sqlBuff.toString());
      System.out.println(sqlBuff.toString());
      Map result = new HashMap();
      while (rs.next()) {
        int id = rs.getInt("ID");
        String name = rs.getString("NAME");
        String descr = rs.getString("DESCR");
        int views = rs.getInt("VIEWS");
        Job job = (Job) mJobMap.get(String.valueOf(id));
        //Populate instance of the Job with data retrieved from database
        job.mName = name;
        job.mDescr = descr;
        job.mViews = views;
      }
      String sqlViewStr = "update trentry set views = views+1 where id in " +
          whereClause;
      // The same trick for Update statement
      stmt.executeUpdate(sqlViewStr);
    }
    catch (SQLException ex) {
      isError = true;
    }
    finally {
      try {

        if (stmt != null) {
          stmt.close();
        }
        if (connection != null) {
          if (isError) {
            connection.rollback();

          }
          else {
            connection.commit();
          }
          connection.close();
        }
      }
      catch (SQLException ex1) { isError = true; }
    }
    FinishJobs(isError);
    return;
  }

  private StringBuffer CreateWhereClause() {
    StringBuffer clause = new StringBuffer("(");
    for (int i = 0; i < mJobs.size(); i++) {
      Job job = (Job) mJobs.get(i);
      String id = job.mID;
      if (i != 0) {
        clause.append(",");
      }
      clause.append(id);
      mJobMap.put(id, job);
    }
    clause.append(")");
    return clause;
  }

  private void FinishJobs(boolean isError) {
    for (int i = 0; i < mJobs.size(); i++) {
      Job job = (Job) mJobs.get(i);
      if (isError) {
        job.mHasFailed=true; //Rudimentary error handling
      }
      job.mJobThread.interrupt();
     /* Wake up the TrainServlet to deliver the page to the browser */
    }
  }

  public void run() {
    Process();
  }
}

　　Worker类创建SQL语句，为一次与数据库的交互来组合Job，用来自数据库的信息生成Job实例并且唤醒Job线程。