我正在运行这个hadoop程序，这是减速机：

public static class joinsReduce extends Reducer<Text, Text, Text, Text>
{   //start class reduce (1)

  public void reduce(Text key, Iterable<Text> values, Context context)
          throws IOException, InterruptedException
  { // start method reduce (2)

    String result = "";

    Map<String, String> memOf = new HashMap<String, String>();
    Map<String, String> subOrg = new HashMap<String, String>();
    Map<String, String> email = new HashMap<String, String>();

    List<String> studList = new ArrayList<String>();

    String line = "";
    String source = "";
    String sub = "";
    String obj = "";
    String hashKey = "";
    String hashVal = "";
    int count = 0;

    for (Text value : values)
    { //start iterate over values (3)
      line = value.toString();
      String[] parts = line.trim().split(",");

      source = parts[0].trim();
      sub = parts[1].trim();
      obj = parts[2].trim();
      hashKey = sub;
      hashVal = obj;

      if (source.equals("type"))
      {
        studList.add(sub);
      }

      if (source.equals("memberOf"))
      {
        memOf.put(hashKey, hashVal);
      }  // end mem

      if (source.equals("subOrganizationOf"))
      {
        subOrg.put(hashKey, hashVal);
      } // end if sub

      if (source.equals("emailAddress"))
      { //(4)
        email.put(hashKey, hashVal);
      } // end if email (4)
    } //end  reading loop iterating over values (3)

    String y = "";
    String z = "";
    String z1 = "";

    for (String x : studList)

    {                      // (6)
      y = memOf.get(x);
      z = email.get(x);
      z1 = subOrg.get(y);

      if (y != null && z != null && z1 != null)
      {     // (8)
        result = x + ',' + y + ',' + z + ',' + z1;
        context.write(new Text(x), new Text(result));
      } // end inner z  if (8)
    } // end iterating loop                         //(6)

  }//end method (2)

}//end class    (1)

它抛出一个错误：java堆空间达到reduce阶段的92%。我用了9个减速机，我不想增加它们。
有没有一种方法可以减少java内存的使用，代码方面？或者可以增加java堆的大小以超过我的主内存（4gb）。
我现在用的是第四个

<property>
    <name>mapreduce.map.java.opts</name>
        <value>-Xmx3072m</value>
    <description>No description</description>
</property>

如果我可以超出可用内存，并且如果我将-xmx3072m更改为-xmx4096m，我该怎么做才能使它生效？