Post reply

General Orientation

  • August 16, 2005 at 10:48 am

    #114843

    I hate to admit that I was just introduced to the concept of Analysis Services at the SQL Roadshow last week.  It sounds very interesting and I think I would like to start using it.  I have found lots of nifty tutorials to help me click here and there.  What I need is help getting my mind around what it is this does.  I can't create my own cubes because I still don't really get what a cube does.  Can anyone point me in a direction to take a look at something that would help me with the conception of this?

    I am happy to use the example cubes provided with Analysis Services but I can't seem to find instructions on what to do with them once you have created them even!

    Thanks,

    Eva

  • August 17, 2005 at 4:35 am

    #582718

    Hi Eva,

    Olap is mainly used for analysis.

    Let us say i have a FMCG Company.

    I would like to know at what place what stocks are there.

    And also you would want to create a report which measures stock against your requirement.

    Or you want to know sales data.

    Let us say you have a hireacrchy that is your company has been divided into various areas under each are you have a store.

    So now you can analyse at various levels for various time periods.

    Let us say for an area for a specific period.

    All this info can be got from OLTP but it is faster in olap.

    The front end you can use is Excel.Using pivot table and connecting to external data source you can connect to these cubes and analyse.

    Elser you can use a third party olap tool.

  • August 17, 2005 at 6:35 am

    #582728

    Take a look at this article, as this explained and clarified my concepts as well:

    http://www.devx.com/dbzone/Article/21410/0/page/1

    Happy learning!

  • August 17, 2005 at 8:44 am

    #582765

    Cubes are analytical tools. They answer simple questions like how many patients we saw on Tuesday at 8:00. Or how many widgets we sold. The idea is to get info into the hands of more people to make better decisions. Many people refer to this as BI or business intelligence, I prefer to refer to it as informed decision making as I think we need to extend it past the "business" paradigm. I have built cubes to look at emergency medical services to see patterns (in injuries, causes, and treatments), look at when we need to bump up staffing and when it can be reduced, as well as traffic accident info. I have a web based example at : http://www.utcodes.org/queryabledata/cube/default.htm  feel free to query away. 

  • August 17, 2005 at 9:19 am

    #582779

    Thank you everyone for the responses.

    Special thanks to Enthusiast for the working link.  I think I needed that to really see what the thing can be used for.

    So, just for a bit more clarification...  Right now I send my users huge grids in Excel that I produce for them from queries based on their requirements.  I put that information into pivot tables for them so the can actually see the results in numbered format, etc.  This feels like it does the same thing.  Am I basically correct in that?

    If I am, my new issue is I am not sure how to select a "fact" table and then which tables get to be "dimensions".  In almost all of the pivot tables I create now I am counting records and that is the aggregate I need.  I do joins of all kinds.  Would the "fact" table be the one in my FROM statement and the "dimension" tables be the ones I have joined, for example?

    Eva

  • August 17, 2005 at 9:45 am

    #582795

    Correct, your giving them the capability to make their own pivot tables.

    The fact table is where the meat of the cube is. It should contain the things you want to count or sum (# of patients, charges, length of stay...) and it should contain the foreign keys to join to your dimensions. The dimensions could be you lookup values. For example I have a Sex (M/F) indicator. In my fact table I store it as 1 or 2. In my Sex dimension table I would have 1 = Male, 2 = Female , 3= Unknown, 4= Indeterminate (just kidding)  and I would join my fact table with my dimension and return the text value Male/Female/Unknow instead of 1/2/3/4. Dates are another great example of dimensions. In my fact table I would store the date of treatment. In my dimension table I could have a list of all dates, what fiscal qtr they correspond to, what day of the week that day fell on, if it was a holiday...

    The dimensions should yield more information about something stored in the fact table. Does that make sense?

    Here is another great tutorial that may be of use http://www.databasejournal.com/features/mssql/article.php/1429671

Viewing 6 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic. Login to reply