登入
 
首頁文件庫學習園地下載支援社群論壇
IT 專業人員的技術資源 >
> 
> 
Design Question

Answered Design Question

  • 2012年4月24日 下午 12:06
     
     
    登入以投票
    0

    Hi,

    I have multiple sources of contacts. Users want to intersect across the multiple sources and also get contacts from both of the sources. For example:

    Query 1. Give me everybody that exist in Source 1 or Source 2.

    Query 2. Give me everybody that exist in Source 1 and also in Source2.

    In Query 2 they want the intersection between the contacts in source 1 and source2.

    Question is should I be creating a Single Dimension of Contacts for Source 1 and Source 2 and create named set for Query 2 ? Or what is the best approach to follow ?

    Thanks,

     

所有回覆

  • 2012年4月24日 下午 02:06
     
     已答覆
    登入以投票
    0

    In my point of view you should go for a single dimension only. By the way do you need to show the source also while reporting ? If not needed while loading the data into the dimension , you will be loading the data from source 1 and from source 2 which is not available in source 1. In that way that should answer all your queries.

    • 已標示為解答 nadirsql 2012年4月30日 上午 01:20
    •  
     
  • 2012年4月24日 下午 06:22
     
     
    登入以投票
    0

    Thanks for the reply.

    Yes users do want to see which source the data belongs to. This DW will be used also for Marketing purpose where the users will be querying the cube to see the contacts which exist in Source 1 but not in Source 2 etc..

    So, in that case, are you suggesting a single dimension for all the contacts and custom named sets for the queries where users want to slice and dice between Source 1 and Source 2 ?

    • 已標示為解答 nadirsql 2012年4月30日 上午 01:20
    • 已取消標示為解答 nadirsql 2012年4月30日 上午 01:20
    •  
     
  • 2012年4月25日 下午 08:25
     
     已答覆
    登入以投票
    0

    I would create a single dimension with one of two options.  If you have a limited number of sources, creating separate columns for each source allows for the simplest of the query options.  This is a method I have used quite often and would recommend it.

    If you have multiple sources with an unkown number or a number that will grow, you might be better served by creating a dimension for sources and  a dimension for contacts and essentially a bridge table between them. 

    Regards, Steve @dataonwheels http://www.dataonwheels.com

    • 已標示為解答 nadirsql 2012年4月30日 上午 01:20
    •  
     
  • 2012年4月27日 下午 12:18
     
     已答覆
    登入以投票
    0

    I wouls suggest a single dimension if the Column set for both the sources is one and the same. We can have the datasource Name to identify the source of the contacts.

    If the column set of both the sources is different its advisable to go for two dimensions instead of one.

    Thanks,

    Sowjanya

    • 已標示為解答 nadirsql 2012年4月30日 上午 01:20
    •