mdx 非重复计数性能答案

【问题标题】：mdx distinct count performancemdx 非重复计数性能
【发布时间】：2015-06-18 05:43:28
【问题描述】：

我得到一个像下面这样的 MDX 查询，它返回过去 30 天的平均不同计数，它返回正确的结果，但性能不好，在我的事实表中每天将有近 2,500,000 行数据。

WITH Member Measures.DailyAverageUser AS
    Avg ( EXISTING{[Date].[Date].[Date].Members},
                 [Measures].[Active Tiles Employee Id Distinct Count]
                ),format_String="##.000000"
             SELECT   Measures.DailyAverageUser ON 0,
            NON EMPTY([Action Targets].[Name].Children)  ON 1
            FROM [OLAP Pre]
            WHERE (
                {[Target Types].[Name].&[tile]},{[Employee Statuses].[Status Id].&      [1],[Employee Statuses].[Status Id].&[3],[Employee Statuses].[Status Id].&[4]},
            {[Business Region].[Abbreviation].&[NONE],[Business Region].[Abbreviation].&[AMS],[Business Region].[Abbreviation].&[APJ],[Business Region].[Abbreviation].&[EMEA]},{[Employee Types].[Bits].&[1],[Employee Types].[Bits].&[5],[Employee Types].[Bits].&[9],[Employee Types].[Bits].&[25],[Employee Types].[Bits].&[13],[Employee Types].[Bits].&[29]},{[Date].[Date Key].&[20150428]:null})

我把查询改成这个了，还是一样的问题，好在这个会缓存20150409：null之间的数据，用20150410：null查询的时候，会很快。

WITH Member Measures.DailyAverageUser AS
             Avg ( {[Date].[Date Key].&[20150409]:null},
                 [Measures].[Active Tiles Employee Id Distinct Count]
                ),format_String="##.000000"
             SELECT   Measures.DailyAverageUser ON 0,
            NON EMPTY([Action Targets].[Name].Children)  ON 1
            FROM [OLAP Pre]
            WHERE (
            {[Target Types].[Name].&[tile]},{[Employee Statuses].[Status Id].&[1],[Employee Statuses].[Status Id].&[3],[Employee Statuses].[Status Id].&[4]},
            {[Business Region].[Abbreviation].&[NONE],[Business Region].[Abbreviation].&[AMS],[Business Region].[Abbreviation].&[APJ],[Business Region].[Abbreviation].&[EMEA]},
            {[Employee Types].[Bits].&[1],[Employee Types].[Bits].&[5],[Employee Types].[Bits].&[9],[Employee Types].[Bits].&[25],[Employee Types].[Bits].&[13],
            [Employee Types].[Bits].&[29]})

需要关于性能的建议

【问题讨论】：

标签： performance count distinct mdx

【解决方案1】：

您想将所有内容都保留在切片器轴上吗？如果没有，您可以像这样将所有内容都放入您成员的定义中 -

WITH Member Measures.DailyAverageUser AS
             Avg (           
             NonEmpty(
                        (
                        {[Target Types].[Name].&[tile]},
                        {[Employee Statuses].[Status Id].&[1],
                         [Employee Statuses].[Status Id].&[3],
                         [Employee Statuses].[Status Id].&[4]},
                        {[Business Region].[Abbreviation].&[NONE],
                        [Business Region].[Abbreviation].&[AMS],
                        [Business Region].[Abbreviation].&[APJ],
                        [Business Region].[Abbreviation].&[EMEA]},
                        {[Employee Types].[Bits].&[1],
                        [Employee Types].[Bits].&[5]
                        ,[Employee Types].[Bits].&[9]
                        ,[Employee Types].[Bits].&[25]
                        ,[Employee Types].[Bits].&[13]
                        ,[Employee Types].[Bits].&[29]},
                        {[Date].[Date Key].&[20150428]:null}
                        ), [Measures].[Active Tiles Employee Id Distinct Count]
                    )            
             ,
                 [Measures].[Active Tiles Employee Id Distinct Count]
                ),format_String="##.000000"

SELECT   Measures.DailyAverageUser ON 0,
NON EMPTY([Action Targets].[Name].Children)  ON 1
FROM [OLAP Pre]

【讨论】：

【解决方案2】：

不知道这是否也很慢？：

SELECT   
  [Measures].[Active Tiles Employee Id Distinct Count] ON 0,
  NON EMPTY
  [Action Targets].[Name].Children  ON 1
FROM [OLAP Pre];

OLAP 可能会因大量不同的计数而变慢。

【讨论】：