Changing your outline isn't just a text update—it's a physical renovation of your data. Database Restructuring is the process of rebuilding the 'House' to fit new rooms.
1. The Definition: Physical Reconciliation
When you change the structure (Dimensions/Members) of a cube, Essbase must reorganize the data on the disk.
* Outline-Only: Renaming a member. It's like changing a sign on a door. (Instant)
* Sparse Restructure: Moving data blocks around. Like moving furniture between rooms. (Fast)
* Dense Restructure: Changing the size of every data block. Like adding a new bathroom to every bedroom in a hotel. (Slow and Resource Heavy)
2. The "House Renovation" Example
Think of your database as a house:
* The Paint Job: Renaming 'Office' to 'Workroom'. No walls moved.
* The Renovation (Dense): You want to add a new 'Account' to a Dense dimension. This is like adding a new closet to every single room in your house. You have to break the walls of every room to make it fit.
Architect Warning: During a 'Dense Restructure', you need enough empty space on your server to hold two copies of the database at once!
3. How it is useful while building an application
* Planning: You learn to avoid making Dense changes during the 'Live' hours of a budget cycle.
* Performance: Minimizing Dense restructures keeps your application healthy and your 'Refresh' times short.
* Disk Management: It prevents 'Server Full' errors by helping you calculate the required temporary space.
4. Where do we use these concepts?
* Refresh Database: In FCCS, Planning, or Essbase Jet UI, this is the button that triggers the restructure.
* Log Files: Checking the 'Application Log' to see if a restructure was 'Dense' or 'Sparse'.
* Outline Design: Deciding to keep a dimension 'Sparse' specifically so that adding members doesn't take 5 hours.
5. Extra Information: The Data Export Rule
Architect Pro-Tip: Before any major restructure, always export your data. If you accidentally change a member's parent or move it to a different part of the tree, Essbase might not be able to find the old data 'address', leading to data loss. Better safe than sorry!