Today, I re-read this best practices article on MSDN – Scale-Out Querying with Analysis Services
The architecture they talk about makes sense. Multiple web servers on the front end, a “query” layer, and processing layer. One thing that I don’t get though is how a setup like that can only support 30-60 concurrent users. I guess I would like to see the numbers on the average concurrent users based on usage. Like how many users on the reports/applications cause concurrent usage? Are connections reused? Etc.
Seems that there could also be a better way than robocopying files around and starting stopping services, but maybe not. Looks like they have an SSIS package that kind of handles it all, which is cool.
Sometimes I wonder though, if these setups are also biased in such a way to maximized the number of SQL Servers (and licenses) that you need to get something like this going. Would be nice to see a study on something like this from a 3rd party, for sure.
Steve Novoselac's Blog 