content_views" class="htmledit_views">

在一些情况中࿰c;可以通过采用已知的功能依赖优势来在数据处理过程中class="tags" href="/tags/YouHua.html" title=优化>优化性能。这种依赖是典型的与产生数据相关的业务规则之结果࿰c;通常不能通过数据本身而推断。

在Mondrian中࿰c;功能依赖(Functional dependency)使用 元素的 属性以及 元素的 属性声明。

property成员的dependsOnLevelValue属性用于指明该成员属性的值功能上依赖哪个成员属性的Level值。换句话说࿰c;对一个给定的leve值࿰c;属性值是稳定的。

Hierarchy中的uniqueKeyLevelName属性用于指出给定的level(如果存在)采用所有在层次活动中更高level作为唯一替代key࿰c;确保对于那些level值的联合是唯一的。

要解释这些内容࿰c;看下面的例子：

cellspacing="0" cellpadding="2" width="788">

color="#ff0000">uniqueKeyLevelName=”color="#0000ff">Vehicle Identification Namber”>                     color="#ff0000">dependsOnLevelValue=”true”/>         color="#ff0000">dependsOnLevelValue=”true”/>     Vehicle Identification Number” column=”vehicle_id” type=”Numeric”/>         color="#ff0000">dependsOnLevelValue=”true” />         color="#ff0000">dependsOnLevelValue=”true” />             color="#ff0000">dependsOnLevelValue=”true” />

在上面的例子中࿰c;我们知道一个给定的制造厂仅存在于一个单独的城市和省࿰c;一个给定的汽车只有一种颜色方案和一种内饰程度࿰c;并且许可证号码与每个单独的省关联。因此࿰c;我们可以说所有这些成员属性功能依赖相关的leve值。

另外࿰c;我们知道车辆识别号码唯一标识每辆车࿰c;每辆车只有一个许可证。因而࿰c;我们得知Make,Model,Manufacturing Plant的联合࿰c;以及Vechicle Identification Number唯一识别每辆车࿰c;license number是冗余的。

这些属性保证Mondrian生成的SQL语句中GROUP BY 能被class="tags" href="/tags/YouHua.html" title=优化>优化。如果缺少任何功能依赖信息࿰c;典型的查询看起来会像下面那样：

cellspacing="0" cellpadding="2" width="600">

ckquote> <code> SELECT `automotive_dim`.`make_id` AS c0, `automotive_dim`.`model_id` AS c1, `automotive_dim`.`plant_id` AS c2, `automotive_dim`.`plant_state_id` AS c3, `automotive_dim`.`plant_city_id` AS c4, `automotive_dim`.`vehicle_id` AS c5, `automotive_dim`.`color_id` AS c6, `automotive_dim`.`trim_id` AS c7, `automotive_dim`.`license_id` AS c8, `automotive_dim`.`license_state_id` AS c9 FROM `automotive_dim` AS `automotive_dim`, GROUP BY `automotive_dim`.`make_id`, `automotive_dim`.`model_id`, `automotive_dim`.`plant_id`, `automotive_dim`.`plant_state_id`, `automotive_dim`.`plant_city_id`, `automotive_dim`.`vehicle_id`, `automotive_dim`.`color_id`, `automotive_dim`.`trim_id`, `automotive_dim`.`license_id`, `automotive_dim`.`license_state_id` ORDER BY `... code> ckquote>

我们知道当在查询中的所有属性也功能依赖于他们的level时࿰c;GROUP BY语句会被认为是冗余的࿰c;并且会完全去掉࿰c;以减少在一些数据库上的性能开销。在上面的例子中࿰c;如果给出了功能依赖上面的SQL语句会变成：

cellspacing="0" cellpadding="2" width="600">

<code>code> SELECT `automotive_dim`.`make_id` AS c0, `automotive_dim`.`model_id` AS c1, `automotive_dim`.`plant_id` AS c2, `automotive_dim`.`plant_state_id` AS c3, `automotive_dim`.`plant_city_id` AS c4, `automotive_dim`.`vehicle_id` AS c5, `automotive_dim`.`color_id` AS c6, `automotive_dim`.`trim_id` AS c7, `automotive_dim`.`license_id` AS c8, `automotive_dim`.`license_state_id` AS c9 FROM `automotive_dim` AS `automotive_dim`, ORDER BY `...

如果查询不够深入到包含unique key的程度࿰c;或者有一些成员属性没有功能依赖于他们的level࿰c;一些数据库（尤其是MYSQL）允许在select中的列不全部出现在GROUP BY 语句中。在这样的一些数据库上࿰c;Mondrian能将功能依赖的属性移出GROUP BY语句࿰c;着眼工会减少SQL查询的处理时间。

cellspacing="0" cellpadding="2" width="600">

<code>code> SELECT `automotive_dim`.`make_id` AS c0, `automotive_dim`.`model_id` AS c1, `automotive_dim`.`plant_id` AS c2, `automotive_dim`.`plant_state_id` AS c3, `automotive_dim`.`plant_city_id` AS c4, `automotive_dim`.`vehicle_id` AS c5, `automotive_dim`.`color_id` AS c6, `automotive_dim`.`trim_id` AS c7, `automotive_dim`.`license_id` AS c8, `automotive_dim`.`license_state_id` AS c9 FROM `automotive_dim` AS `automotive_dim`, GROUP BY `automotive_dim`.`make_id`, `automotive_dim`.`model_id`, `automotive_dim`.`plant_id`, `automotive_dim`.`vehicle_id`, `automotive_dim`.`license_id`, ORDER BY `...

注意࿰c;在Mondrian4.0时可能会有所改变。