[TIP 115] - In-Memory Data Cubes to Enhance Performance

Are you using In-Memory Data Cubes to help enhance your performance? If not, you should be!

One potential problem with In-Memory cubes is that they can take a lot of resources from a CPU perspective to build. What if you have cubes that need to be built during the day, while your users are still accessing Dundas BI? You probably don’t want to see performance degradation because your cube is building behind the scenes.

Try this advanced setting in Dundas BI to allow you to change the number of CPU cores to be used when building a cube. A good idea might be to use half of your available CPU cores so that your users can still have the other half dedicated to the performance of the application:


Has anyone played with this? We’re seeing a phenomenon where Dundas seems to lean heavily on one core of our server, leaving the others relatively-idle. This, of course, has performance hits.

Running Postgres db on Ubuntu servers, if you’re curious