Users normally complain about their login times and how snappy the OS is, which are the two normal user metrics we have to deal with it. Used LoginVSI since it is the industry standard and so people could reproduce their results. Used LoginVSI 4.0 since it still had the office document test because some vendors would use office 2003 instead of 2010 or 2013 because the medium workload would look better. They used UberAgent with Splunk to measure more things outside of what LoginVSI would miss.
XenApp 6.5 vs XenApp 7.5:
Took a hit going from XA65 to XA75 on the same 2008 R2 image.
7.5 on 2012 was faster than the base. (2012 R2 was barely better)
2008 R2 2012, 2012 R2:
New base was faster with 2012 R2 over 2008 R2, and 2012 was just a little slower.
10GBe vs 1GBe:
Didn’t make a huge difference with profile load.
SMB vs 2.1 vs SMB 3.0:
SMB 3.0 was slower for some reason and will require further testing.
Normal vs Accelerated Storage:
Normal storage made it twice as slow with normal storage.
Old hardware vs New hardware:
Adjusted Power Settings on the physical servers to balanced you always need to change this to Max Performance. Power balanced took at least 33% hit over maximum performance.
First base test 2008 R2 xa65 accelerated storage SMB 2.1 with 1GB
Second test 2012 R2 xa75 accelerated storage SMB 3 with 10GB,,, the first base and new base was a drastic change about 2-3 times faster which was 4 years newer.
I know I cannot wait to see the more detailed results, this is a great set of tests and they are planning even more tests with Moonshot, GPU, XenDesktop, Further SMB3 test and look at App-V shared content store.