Great, looking forward to it. If you have any considerations concerning processor selection for a SSAS-only server, I would appreciate, if you could share them
]]>I meant not the benchmark itself but rather some comparison of processors similar to your blog post but with focus on DW/OLAP workloads
]]>The TPC-H benchmark is for DW workloads.
]]>do you know whether there is a similar analysis for DWH/OLAP workloads?
]]>Depending when they did this, it was probably not the best idea for performance and scalability. A two-socket Nehalem-EP or newer server would be much faster than an equivalent four-socket server (that was populated with only two processors).
This changes somewhat with the latest Intel E7-4800 v2 family, where the processors are much closer to the Xeon E5-2600 v2 family in performance.
]]>The extra CPUs were reserved on the off chance that they’d become necessary for performance reasons, albeit with the corresponding licensing hit. I can’t remember a time that they were installed before the server itself was replaced.
That was more common back in the Xeon X7460/X7560 eras than it is now, with there now being so much diversity in Intel’s Xeon line.
]]>