我的做法是先在schema中定义好聚合表的相关值 ，如：

<cube name="cube_agent_state" caption="test" encoding="UTF-8">

<Table name="fact_agent_status_fact">

<AggName name="agg_c_fact_agent_status_fact" >

<AggFactCount column="fact_count"/>

<AggForeignKey factColumn="time_id" aggColumn="time_id" />

<AggForeignKey factColumn="agent_no" aggColumn="agent_no" />

<AggForeignKey factColumn="agent_status" aggColumn="agent_status" />

<AggForeignKey factColumn="reason_code" aggColumn="reason_code" />

<AggMeasure name="[Measures].[times]" column="times"/>

<AggMeasure name="[Measures].[length]" column="length"/>

</AggName>

</Table>

<Dimension name="times" foreignKey="time_id" caption="时间">

<Hierarchy hasAll="true" primaryKey="time_id" >

<Table name="dim_date" />

<Level name="year" column="y_id" uniqueMembers="false" type="String" levelType="TimeYears" caption="年"/>

<Level name="month" column="m_id" uniqueMembers="false" type="String" levelType="TimeMonths" caption="月"/>

<Level name="week" column="w_b_id" uniqueMembers="false" type="String" levelType="TimeWeeks" caption="周"/>

<Level name="day" column="d_id" uniqueMembers="false" type="String" levelType="TimeDays" caption="日"/>

<Level name="minute" column="begin_time" uniqueMembers="false" type="String" levelType="TimeMinutes" caption="开始时间"/>

</Hierarchy>

</Dimension>

<Dimension name="agent_nos" foreignKey="agent_no" caption="座席编号">

<Hierarchy hasAll="true" allMemberName="agent_no" primaryKey="no" primaryKeyTable="dim_agent">

<Table name="dim_agent" />

<Level name="agent_no" column="no" table="dim_agent" caption="座席编号"/>

</Hierarchy>

</Dimension>

<Dimension name="agent_status" foreignKey="agent_status" caption="座席状态">

<Hierarchy hasAll="true" allMemberName="agent_status" primaryKey="agent_status" >

<Table name="dim_agent_status" />

<Level name="agent_status_name" column="agent_status_name" uniqueMembers="false" levelType="Regular" type="String" caption="座席状态"/>

</Hierarchy>

</Dimension>

<Dimension name="reason_code" foreignKey="reason_code" caption="小休类型" >

<Hierarchy hasAll="true" allMemberName="reason_code" primaryKey="value" >

<Table name="dim_dict" />

<Level name="reason_code_name" column="name" uniqueMembers="false" levelType="Regular" type="String" primaryKeyTable="dim_dict" caption="小休原因"/>

</Hierarchy>

</Dimension>

<Measure name="length" column="length" aggregator="sum" formatString="#,###" datatype="Numeric" caption="总时长"/>

<Measure name="times" column="id" aggregator="count" formatString="#,###" datatype="Numeric" caption="次数"/>

</cube>

然后再建立聚合表

DROP TABLE IF EXISTS `agg_c_fact_agent_status`;

CREATE TABLE `agg_c_fact_agent_status_fact` (

`fact_count` int(11) DEFAULT NULL,

`time_id` varchar(50) DEFAULT NULL,

`agent_no` varchar(20) DEFAULT NULL,

`agent_status` int(1) DEFAULT NULL,

`reason_code` char(1) DEFAULT NULL,

`times` int(11) DEFAULT NULL,

`length` int(11) DEFAULT NULL,

`report_time` datetime DEFAULT NULL,

KEY `i_fasf_time_id` (`time_id`),

KEY `i_fasf_agent_no` (`agent_no`),

KEY `i_fasf_agent_status` (`agent_status`),

KEY `i_fasf_reason_code` (`reason_code`)

) ENGINE=InnoDB DEFAULT CHARSET=utf8;

特别说明：AggMeasure值是必须与原schema中定义的名称是相同的,如果measure是CalculatedMember类型，那么就不必要建立在聚合表中，否则会报错，在实际使用中，会自动使用因为这个是在Mondrian引擎中处理的，所以不需要处理