So everyone knows that Attributes can be an awesome feature to add to Essbase cube. I use them all the time. A primary example would be your Customer dimension in Essbase. It's a 'stored' dimension, but you need to include many facets of the the customer...like Country/State, Size, Sales Rep, etc.
Trying to put all these components into the Customer hierarchy could become very cumbersome. In come Attribute dimensions. What's great about them is your don't have to load any data to them, you can retrieve from them directly (since the are aligned with the Customer dimension).
I had a client that had 30+ dimensions with 5 attribute dimensions and retrieval performance started to take a hit. These guys especially loved to drill down by most (if not all dimensions), so they asked what's the best approach for doing that type of retrieval. First of all, I don't recommend it. But hey, they are my clients and I will assist them however they need it.
One approach would be to but the attributes in the right most columns before the data for faster retrieval while keeping the stored dimensions on the left. I realize that may not give the "look" the user what hoping for, but it will return faster retrieval.
No comments:
Post a Comment