Re: OLAP Cube Testing
From: Arindam Banerjee (ArindamBanerjee_at_discussions.microsoft.com)
Date: 01/19/05
- Next message: Steve McHugh: "Re: AS Enterprise features not avilable"
- Previous message: Muhammad Mansoor: "MDX Problem"
- In reply to: Rich Millman: "Re: OLAP Cube Testing"
- Messages sorted by: [ date ] [ thread ]
Date: Wed, 19 Jan 2005 00:43:01 -0800
Hi Rich,
Thanks for your response. To be specific about the project, I have 30 Dims
and 4 facts. It is taking the inventory data which is comming from SAP and
other forecasting tool, staged and then through a star schema it is
populating into the cube through DTS.
I have kept check for the data in a manner of count (*) and NOT IN type of
queries. Since i am a black box tester, the scope is very limited for me. Two
facts are joining to build the cube and it is displayed as Perspective to the
end users. In the perspective also, I am checking the aggregations as well as
the normal data checking. i can validate the data in frontend(Perspective) as
well as in backend(SQL Server). Is there any other way to check for more?
What are the critical areas where data loss could occur?
Thanks,
Arindam.
P.S.: It will be a nice idea to provide your e-mail alias, mine is:
v-ariban@microsoft.com
"Rich Millman" wrote:
> OK, this topic could fill a couple of books, but I'll focus on some of the
> key points. Since you say that nobody in your company knows anything about
> cube building/testing, I will start from the beginning.
>
> An OLAP cube is a mechanism to define (and optionally store) aggregations.
> At it's lowest level, it aggregates data (facts or measures) according to
> the various dimensions provided. An example of a fact would be sales, while
> dimensions are what put your facts into context, such as product or time.
>
> Cubes can do so much more than just aggregation, if you provide special
> rules. I won't get into that here.
>
> I assume that someone is providing the cube for you. In order to test the
> cube, you must understand the business rules that went into the design and
> creation of the cube. If the business rules are simple aggregation (sum),
> then we work with the following:
>
> In the relational database that provided the source for the cube, there is a
> fact table and optionally dimension tables. The fact table must tie back to
> the dimension tables. You should be able to generate a relational query
> that returns the same results as your cube (as long as you are only
> summing). A sample SQL Query would be:
> Select sum(sales) from sales_table where product = 'turkey'
>
> The above is a very simple representation, but with it, you then have one
> dimension (it is assumed that the sales table has a product dimension in
> it). And can get a total value. That value should match the value obtained
> by the cube.
>
> (Note everything mentioned above is for the simplest example, it could be a
> whole lot different if you are doing complex equations or overrides within
> the cube)
>
> Tools to test the cube for data (in your situation) would be Excel, by
> attaching to the cube and dragging and dropping the information into a pivot
> table, or Analysis Server Manager. Neither of these tools requires you to
> know MDX in order to return the data.
>
> Hope this helps. If you need anything more specific, post again for help
> and I'll send my address.
> "Arindam Banerjee" <Arindam Banerjee@discussions.microsoft.com> wrote in
> message news:1BC36D9B-286B-430D-A39A-AA7FFFF285F7@microsoft.com...
> > Hi Folks,
> > this is my first mail in this site. I am a tester and I need to test an
> > OLAP
> > Cube. Nobody in our company knows anything about Cube Building and
> > Testing.
> > Could anybody tell me how to test a cube and what are the different
> > critical
> > areas of testing, I will be very obliged.
> > Thanks,
> > Arindam
> >
> > arindam2004@yahoo.com
>
>
>
- Next message: Steve McHugh: "Re: AS Enterprise features not avilable"
- Previous message: Muhammad Mansoor: "MDX Problem"
- In reply to: Rich Millman: "Re: OLAP Cube Testing"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|