December 9, 2017 at 8:20 am
All,
We are to do a one-time initial load of our SSAS Tabular Cube and then perform incremental updates also using CDC. This one-time initial load is failing because of memory issues. We had an outside company develop most of this code(i.e. Loading reporting database staging, then loading the reporting edw, then loading the SSAS Cube). This one time run is really eating up all of our memory. I was able to look at the SSIS package that loads the reporting databases and they are using the lookup task. I noticed that they are trying to load the tables into memory even though they are only mapping a few columns. I told them that they shouldn't be loading the tables into memory just the columns that they really need. In some cases loading the entire table when it's a small table isn't and issue rather not a best practice. I've examined the memory settings in SSAS and they appear to be fine. The database enging has a max memory limit but not a minimum memory limit so SSAS should be able to get as much memory as it needs. I probably need to set the minimum memory limit on the database engine to a value cause SSAS will bring the database engine to its knees if I don't. We have 180 gb's of memory and I've set the parameter in SSAS to utilize paging if it needs to. Unfortunately I haven't had the opportunity to see if the internal processes of all of the SSIS Packages and the load of the cub process is as efficient as it needs to be yet. Any recommendations would be greatly appreciated. By the way the server that this is running on has mutliple instances(i.e. Database Engine, SSIS, SSAS, SSRS) running.
Viewing 0 posts
You must be logged in to reply to this topic. Login to reply