Anyone using data-mart function ?

Everything is in the title 

I'm wondering what is the intended use of this function and if many people are using it ? And for what objective ?

 

Thanks,

Etienne

Answers

  • Hi Etienne CAUSSE,

     

    you might want to have a look at this document Cluster  describing the need for datamart in certain circumstances.

     

    Hope, this helps!

     

    KR
    Helmut

  • Hi Etienne CAUSSE,

     

    I did once discuss the idea (but did not implement!) of distributing child databases to a sales team for offline use. You could run a master database on your central board server and use the data mart function to create child databases along an entity (e. g. sales representative). These child dbs (containing only their slice of the data) would then be distributed to the user clients. The clients would run their own board environment and would be able to work "offline".

     

    Another use case could be, to quickly create a smaller copy of your database, that only contains a portion of the data (e. g. for testing purposes with only one year of data).

     

    But to really answer your question: besides showing the function in trainings I am not using it :-)

     

    BR

     

    Sebastian

  • Hi Helmut Heimann,

    I remember reading that article some time ago, but it seems only about splitting a big database in several machines, which is not required in our case.

    Did you already use the datamart function on the same server ?

  • Hi Sebastian Gurt,

    If I understand correctly, this would allow to synchronise the content of a DB in a separate one, potentially in the same server ?

    I have been wondering if this kind of use could help us in managing some contradictory access rights that might be requested, as mentioned in other topics.

    A copy of a single slice in a separate DB could allow access through a different DB profile, for the same user ?

     

    Thanks for your input.

    Etienne

  • I wouldn't call it synchronization as it only works in one direction (it is only a copy from the original db with the different options to create slices by time, by entity, by user profile, ...).

    Etienne CAUSSE wrote:

    I have been wondering if this kind of use could help us in managing some contradictory access rights that might be requested, as mentioned in other topics.

    A copy of a single slice in a separate DB could allow access through a different DB profile, for the same user ?

    I did not try or even think about something like that yet - but it sounds interesting. In theory it should work. As the datamart function creates a new database with a new name you could have a different db profile assigned for this database.

  • I'm ok with synchronisation one-way only, in that case it is almost a pure BI database, no data entry for the moment.

    Is there a procedure somewhere to explain how to setup the datamart on the same server ?

  • Sebastian Gurt
    Sebastian Gurt Employee
    First Anniversary DACH Level 100: Foundations of Building in Board Level 200: Leveraging Board for Business Insights
    edited March 2020

    You just run the procedure action and use the option that fits your needs:

    image

     

    1. This replaces databases with the same name
    2. save database to ...\BOARD\Datamart\ (instead of ...\BOARD\Database\)
    3. create the database with a slice of the time range
    4. create a database copy with the name you put here
    5. create database copies along an entity (e. g. if you have a sales rep entity with 6 entries you get 6 databases with just that slice of data)
    6. create a database copie along a user profile
    7. save the database copies to a custom path

     

     

    Hope this helps. I suggest you just try the different options with a small database on you local machine.

     

    BR

     

    Sebastian

  • Hi Etienne CAUSSE,

    I've discussed that with a customer but with the resulting databases being distributed to different servers, thus implemeting the cluster. In that particular case we decided otherwise. And--just like Sebastian Gurt mentioned--I've only implemented that for test cases on my local machine with a really small database.

     

    KR
    Helmut

  • Thanks for the explanation, you might suggest to the lab to include it directly in the official documentation

    I will do some testing to see if that can help with some of our issues and how it behaves in terms of performance.

  • I opened a ticket to update the online help.