Dimensional Modelling : Online Media Ads Company

View previous topic View next topic Go down

Dimensional Modelling : Online Media Ads Company

Post  kuldeepchitrakar on Wed Apr 21, 2010 3:27 am

Hi

I am designing a Dimensional Model for a company which provides ads to website.

they track the ad performance and load the log files into 3 different tables.

e.g. Impressions log, Click log and Event Log

now these table are pretty huge and we are trying to deign a datawarehouse which would help them to slice and dice data in all manner and also give good performance.

The problem is that we need to have grain at click,impression and event level which makes fact table very huge.

If anybody has faced similar situation she/he can give me a pointers for above situation

What are the best strategies to be followed in such a case while designing DW schema.

This is my first DW project

kuldeepchitrakar

Posts : 17
Join date : 2010-04-21
Age : 34
Location : India

View user profile http://www.bidwbooks.com

Back to top Go down

Re: Dimensional Modelling : Online Media Ads Company

Post  ngalemmo on Wed Apr 21, 2010 12:07 pm

I've done clickstream before. The modeling itself is pretty straight forward. There is typically a user and a session cookie that helps identify basic contexts. Loads are fairly simple since facts are not updated. Stick to a clean dimensional model and pick the right platform to host the data.
avatar
ngalemmo

Posts : 3000
Join date : 2009-05-15
Location : Los Angeles

View user profile http://aginity.com

Back to top Go down

Re: Dimensional Modelling : Online Media Ads Company

Post  kuldeepchitrakar on Thu Apr 22, 2010 3:46 am

Thanks for inputs.

After going through current log table structure.

I could see there are three fact table to be developed.

1. event
2. Click
3. Impression

Event - open page, go to a page etc.
Impression : ad is shown to user
click after seeing ads and he clicks on impression

and since these things will not happen at a time. its good to have three different fact table.

Please comment on my understanding.

Also, how do you guyes tackle Unique Users measure in such a model. as it can not be aggrgated and having Fact table at lowest granularity will raise a performance flag.

kuldeepchitrakar

Posts : 17
Join date : 2010-04-21
Age : 34
Location : India

View user profile http://www.bidwbooks.com

Back to top Go down

Re: Dimensional Modelling : Online Media Ads Company

Post  Sponsored content


Sponsored content


Back to top Go down

View previous topic View next topic Back to top

- Similar topics

 
Permissions in this forum:
You cannot reply to topics in this forum