Question

我最近遇到了一些JPA查询的性能问题，并开始研究解决它的方法。

我有以下命名查询：

<?php session_start();
require_once 'connect.php';
$a = 0;
$conn = @new mysqli($host, $dbuser,$dbpass,$dbname);
if ($conn->connect_errno!=0){
     echo "Error ".$conn->connect_errno;
} else {
     $id=$_SESSION["id"];
     //echo 'to jest nasze id '.$id;
     $sql = "SELECT * FROM zgloszenia WHERE user_id='$id'";
     $wynik = $conn->query($sql);

    echo '<table class="table table-striped">
         <tr>
            <th>id</th>
            <th>stan</th>
            <th>opis</th>
            <th>data</th>
         </tr>';

     while($row = $wynik->fetch_assoc()) {
            echo "<tr onclick='window.location='pagename.php?id= ".$row['id']."''>
                    <td>".$row['id']."</td>         
                    <td>".$row['stan']."</td>
                    <td>".$row['opis']."</td>
                    <td>".$row['data_zgl']."</td>";
            echo "</tr>";
        $a++;
       }
       echo '</table>';
}
print_r($t);
$conn->close();
?>

然后我将其定时如下：

@NamedQuery(name = "EquipmentBooked.findAll", query = "SELECT e FROM EquipmentBooked e")

结果是：

String timings += "\nEquipmentBooked";
long startTime = System.currentTimeMillis();
List<EquipmentBooked> equipmentBookings = entityManager.createNamedQuery("EquipmentBooked.findAll", EquipmentBooked.class).getResultList();
timings += " - Query time [ms]: " + (System.currentTimeMillis() - startTime);
for(EquipmentBooked booking : equipmentBookings){
    log.info(booking.getEquipmentBookedId());
}
timings += " - Total time [ms]: " + (System.currentTimeMillis() - startTime);

333秒仅约2000条记录！

然后我决定试一试本机查询：

EquipmentBooked - Query time [ms]: 333405 - Total time [ms]: 333460

这导致：

String timings += "\nEquipmentBooked (native)";
long startTime = System.currentTimeMillis();
List<EquipmentBooked> nativeEquipmentBooked = entityManager.createNativeQuery("select * from EQUIPMENT_BOOKED", EquipmentBooked.class).getResultList();
timings += " - Query time [ms]: " + (System.currentTimeMillis() - startTime);
for(EquipmentBooked booking : nativeEquipmentBooked){
    log.info(booking.getEquipmentBookedId());
}
timings += " - Total time [ms]: " + (System.currentTimeMillis() - startTime);

6秒！好多了。

当我在同一个方法中运行这两个查询时，有趣的是，本机首先是命名查询。结果是：

EquipmentBooked (native) - Query time [ms]: 6607 - Total time [ms]: 6767

有人可以解释为什么本机查询速度更快，并提高了命名查询的性能吗？

注意：这些测试都是在任何实体被缓存并且数据库是Oracle之前运行的。

更新：删除了联接提取以确保查询相同。

更新2：在查看生成的SQL之后，我能看到的唯一区别如下：

EquipmentBooked (native) - Query time [ms]: 6607 - Total time [ms]: 6767
EquipmentBooked - Query time [ms]: 699 - Total time [ms]: 774

Vs以上。

SELECT EQUIPMENT_BOOKED_ID, CALENDAR_ENTRY_ID, EQUIPMENT_BOOKED_STATUS_ID, EQUIPMENT_ID, EXPERIMENT_PART_ID FROM EQUIPMENT_BOOKED

在这两种情况下，JPA驱动程序都会以相同的方式关闭并获取（许多）关系。 E.g：

select * from EQUIPMENT_BOOKED

为什么本机查询会大大加快命名查询的速度？

0 个答案: