I recently built a cluster of new to me Dell R620s, which run HyperV with various services. I noticed right away that these servers were running slower than our old servers, with poor mysql performance and folder transfers. I then spent the next three weeks tweaking every windows and bios settings I could find, updating everything, and purchasing different Nics, raid controllers, and drives in an attempt to fix it. The problem was mainly with small io intensive tasks. A folder with thousands of small files would copy over on local disks in 23 seconds on my old server, but take 2 min 40 sec on the new servers. Large file transfers were fine however. The fastest way to test and compare the problem was to use crystaldiskmark, and look at the 4k random tests, since sequential tests performed fine. The Q1T1 for example would run 21mb read / 24mb write on the new server, and 35 / 38 on the old server. Finally it occurred to me to fire up an old 2008r2 VM on my 2019 hyper-v host, and to my surprise the number were a good bit better at 25 / 30. So with that, I scrapped 2019 completely for 2016 on the host, and sure enough it ran 74 / 65 right out of the gate.
Now if someone has any ideas of what's wrong or changed in 2019, I'd love to save having to rebuild the entire cluster. Of course the strange thing is that a 2019 VM ran even slower than a 2008r2/2016 VM on a 2019 host, which I would think would eliminate most of the driver issues on the VM level. Hopefully this helps someone out there, and at the very least points out that there are still serious issues with 2019.