Some further feedback. We have pointed Tableau to a smaller Essbase cube on the same server and the response times are quick.
There are two major differences between the two Essbase cubes: The size of the actual cubes and the number of measures.
The cube giving us an issue is 5 GB in size and has approximately 9, 000 stored accounts and 15, 000 dynamic accounts. The Measures list is very large.
We too use an Essbase Cube, and it took me a while to "optimize" the reports I've built.
Our cube is just over 60 gig, and has almost 1 million data points.
Things that I found to be very helpful...
I found that setting dimensions to "context" where appropriate. (rule of thumb is that the cut of the data would be 1/2 of original data set). For example, 95% of my users only work in US dollars, so currency is one that I "add to context"
I use "scenario" as a utility dimension. This means measures can be set to any specific scenario in the database. In our case, that's Actual, Budget, and Forecast.
Lastly, I've learned to only use quick filters where absolutely necessary. Our cube has fairly deep generation count, and therefore I try to minimize the number of times I put a quick filter on a worksheet. That also implies that I set filters to "global" as often as possible.
Lastly, I did find that using "Sets" sped things up slightly, but I didn't like the static nature of the set list. I needed it to be more dynamic.
Hope this helps
you could do two things:
1. download and run Tableau Performance Analyzer provided by certified partner InterWorks.
Troubleshoot Your Load Times
2. ask your account manager / tableau representative to enroll you as beta tester, Tableau 8 have a facility called performance recorder and you can enable it to collect various performance metrics. The final result can be inspected by end users and exported as packaged workbook (.twbx)
The performance workbook is filtered to show only the most important things. But additional details are kept that may indicate the problem, so it still helps to share it with support if you are willing.
Below image shows a performance summary against a test workbook, 130M records, stored on Tableau Data Engine, on my C: vintage laptop drive. The figures of elapsed time on tasks are not important, and should not be treated as a baseline or a benchmark. I provide the print screen just to have an idea how performance summary looks.
Hope this helps.