Real Estate Investment Trust
| Function: | Products: |
| Reporting | IBM Planning Analytics |
Challenge
A publicly traded real estate investment trust (REIT) with more than 3,000 properties in nearly all the U.S. states asked Revelwood to help with an upgrade of their Planning Analytics environment and to convert their existing reports to the new version. During this process, we discovered that a Planning Analytics allocation cube was using a massive amount of RAM, resulting in abnormally slow processing and performance.
Solution
We dug deeper. Why that cube was so large? The company has approximately 1,000 tenants, with some expenses for each property allocated to the associated tenant. The model was originally built to process allocations to every tenant – even in situations where the property only had one tenant. As a result of this, the cube was almost 1,000 times larger than the pre-allocation expense cube. The pre-allocation cube used 40MB of memory and the post-allocation cube was using more than 30GB of memory.
Results
Working with the client, we discovered that almost all of the properties were associated with only one tenant. We created a TurboIntegrator script to determine the applicable property-tenant combinations by analyzing data in one of the allocation definition cubes. If data existed, an assumption was made that the combination existed. We then updated the rules to ensure that the calculations were only processed for the applicable combinations. The end result? We trimmed down a cube that was using more than 30GB to one requiring only 65MB of memory. This was an improvement of more than 99%.