We observe that MTCubeLoad=T parameter does not work for all cubes. Some of the cubes are loaded in parallel using multiple CPUs, and some others are not. Moreover, as per my observation, which cubes are loaded in parallel also changes from time to time. When I check the cubes that are loaded in parallel in different server starts, I see that different cubes are loaded in parallel. Hence I assume it may not be cube specific, rather it might be a defect in determining to load the cube in parallel or not. I can expect small cubes to be loaded in single threads, that it not an issue, but I see this variation in large cubes.
Due to this problem, server load time 1) becomes too long, 2) varies a lot in each restart.
Have anyone else had the same observation? What might be causing this issue?
Thanks in advance for all replies.
1 post • Page 1 of 1