I probably should have mentioned this a while ago, but on Thursday at 8-9pm GMT (which is 3-4pm EST) I’m doing a webinar for the guys at Pragmatic Works, on “Performance Tuning Analysis Services 2008”. It’s going to be a basic-level introduction to the subject, covering topics such as partitioning, aggregations and MDX calculations. You can sign up here:
https://www1.gotomeeting.com/register/824246961
Hello Chris,
I have missed this. do we have a chance to watch
It should be, but I can’t find it. You can see pretty much the same presentation here:
http://content.sqlbits.com/Sessions/Event6/Introduction_to_Performance_Tuning_Analysis_Services_2008
Hi Chris ! Good Day!
I need your inputs/suggestions regarding performance tuning of cubes. Currently I am facing a lot of issues w.r.t performance of SSAS cubes in my project. The design of the cube is pretty simple and straight forward. There are no M2M relationship or reference relationship in my Dimension usage. However we have written a lot of SCOPE statements to meet the business requirements.
My issues are the reports take a lot of time to open up. We are using Panorama for Adhoc analysis and any kind of cross drill on the dimension or any drill down from a higher lever to the lowest level directly takes a lot of time in rendering the data.
I have done my initial bit of analysis and I ran the UBO. I enabled the Query logging for a Month and I ran the UBO and the aggregations were designed as per the suggestion by the wizard. However when I try to process the cube with the new Aggregations, My cubes are failing because the new aggregations are consuming a lot of disk space and I get the File system error stating that there are no enough space on the disk.
Please let me know where am i going wrong. I am now considering to create partitions and then try out the UBO options.
Could you please let me know the best approach in resolving this issue?
Thanks a lot !
Regards,
Ram
Hi Ram,
First question: are the Scopes causing your problems? Try deleting all your aggregations and then commenting out everything in the MDX Script and reprocessing, then run your queries. Is everything fast?
As far as the aggregations go, if you are running out of disk space you either need to delete some aggregations or get more disk space! To be honest it sounds like you’ve built too many aggregations, so I would try rerunning UBO but building no more than 10-20 aggregations (try limiting Performance Gain to around 30%).
Chris