It is amazing how much similarity I found between the subject of this article and common mistake made by most Siebel Analytics professionals in the field. My opinion has always been to improve upon existing ideas or solutions. It is a waist of time and resource to reinvent the wheel.
I have clients calling with performance problems of ETL recurring daily jobs that takes more than 34 hours to complete. Just recently, I had a talk with a a fortune 500 with query performance of its application when the OLAP database is not more than 10 GB in size. Whenever I ask if they have built custom stars instead of trying to re utilize the one that come out of the box. I get an excuse that I have proven so many time to be wrong "Our needs were to specific or our OLTP is heavily customize"
People, you are dealing with a new paradigm now. It is call service oriented architecture (SOA) Stop looking for programmers to implement these applications. They are all moving to the offshore centers so they can write the reusable code of these wonderful applications. What you need is the collaboration of your business and IT teams in your BI efforts. You will also need architects who understand the new technology and are able to interact with your team.
Siebel Analytics Applications answer the who, what, where and when. Who are customers, vendors, employees, partners, etc. What are tangible and intangible products. Where is usually a refer to a geography dimensions or markets where you conduct business. When is your time dimension. It is unreal to believe what this client just told me yesterday "We had to build custom dimension because they were missing from the OOB model" I still thinking what does this client mean by that. I bet you own or have owned a product item from this client. That means that you are or were a customers of it. So why they did not use the W_PRODUCT_D, W_CONTACT_D and W_TIME_D is beyond me. Just take a look on how rich this model is by the picture of the star schema attached to this entry.
I have clients calling with performance problems of ETL recurring daily jobs that takes more than 34 hours to complete. Just recently, I had a talk with a a fortune 500 with query performance of its application when the OLAP database is not more than 10 GB in size. Whenever I ask if they have built custom stars instead of trying to re utilize the one that come out of the box. I get an excuse that I have proven so many time to be wrong "Our needs were to specific or our OLTP is heavily customize"
People, you are dealing with a new paradigm now. It is call service oriented architecture (SOA) Stop looking for programmers to implement these applications. They are all moving to the offshore centers so they can write the reusable code of these wonderful applications. What you need is the collaboration of your business and IT teams in your BI efforts. You will also need architects who understand the new technology and are able to interact with your team.
Siebel Analytics Applications answer the who, what, where and when. Who are customers, vendors, employees, partners, etc. What are tangible and intangible products. Where is usually a refer to a geography dimensions or markets where you conduct business. When is your time dimension. It is unreal to believe what this client just told me yesterday "We had to build custom dimension because they were missing from the OOB model" I still thinking what does this client mean by that. I bet you own or have owned a product item from this client. That means that you are or were a customers of it. So why they did not use the W_PRODUCT_D, W_CONTACT_D and W_TIME_D is beyond me. Just take a look on how rich this model is by the picture of the star schema attached to this entry.
No comments:
Post a Comment