November 7, 2012 at 12:04 am
Comments posted to this topic are about the item The OLAP Sprint
November 7, 2012 at 8:40 am
Good luck! I taught myself OLAP by taking a simplified copy of an existing relational database and reformatting everything to a more dimensional model. Even with a book on OLAP and plenty of web resources, it took me more than a week just to get the data elements correctly formatted so that I could actually use them in SSAS (key structures and all that). If you use the AdventureWorks or Contoso databases though, it isn't nearly as hard since they are already set up in a clean format and the data is pretty much perfect.
November 7, 2012 at 11:47 am
Thanks for the encouragement, KWymore.
It sounds like I'm mimicking your approach as well - taking an existing relational OLTP model and building from that. I've used view in my relational DB to act as my fact and dimension tables, so that I don't have to build a data warehouse and worry about ETL for my proof of concept. Although I know the Contoso and AW sample DBs are comprehensive, I really wanted to work with my data to see it come to life. So far the experience has been great.
November 12, 2012 at 9:24 am
It really started to come together for me when I used my own data too. The data in AW and Contoso is a little too perfect. That data is great for getting used SSAS but that is about it.
I also found that it helps to have a typical business question in mind when designing your cube. You want it flexible enough that you can answer the typical business questions plus some value added points (This way you can justify the need for the cube). If you try to pull in your entire universe of data and then filter it down, you will end up creating a huge mess. A lot of BI projects fail because the tool doesn't answer the important questions. it sounds like you already have an idea of what you want your cube to do though. +
November 14, 2012 at 6:29 am
KWymore (11/12/2012)
It really started to come together for me when I used my own data too. The data in AW and Contoso is a little too perfect. That data is great for getting used SSAS but that is about it.I also found that it helps to have a typical business question in mind when designing your cube. You want it flexible enough that you can answer the typical business questions plus some value added points (This way you can justify the need for the cube). If you try to pull in your entire universe of data and then filter it down, you will end up creating a huge mess. A lot of BI projects fail because the tool doesn't answer the important questions. it sounds like you already have an idea of what you want your cube to do though. +
It was definitely a priority to use my own data, for two main reasons:
1) I wanted one of the benefits of the development of the prototype to be some actual exploration of the data, not just by me, but by the executives and business analysts here, because I knew it would really engage their appetite for pursuing BI beyond the prototype. I also knew this would help with the justification for the conference expenses (registration, travel, lodging).
2) I knew that if I used my data, I'd be able to confirm that the results that the cube was producing were accurate.
"Nothing like sleeping in your own bed", as they say.
November 14, 2012 at 8:33 am
Exactly. We had a hard time selling dashboards to our executives until we decided to just do the work and show them what was possible. Now it is our main focus. Funny how that works, right?
Glad you are getting something out of the PASS Summit. I may have to talk my boss into sending me next year. Keep up the posts!
Viewing 6 posts - 1 through 5 (of 5 total)
You must be logged in to reply to this topic. Login to reply