I had made this production architecture in order to significantly improve the performance.
3 dedicated VMS running SQL server 2017 (sharing 12 cores of Xeon 1670 v2 (3.5Ghz) with 64gb ram and SSD)
vs
1 VMS running MS SQL Server 2014 (sharing 12 cores of Xeon 1670 v2 (3.5Ghz) with 64gb ram and SSD)
Why after deploying everything the performance of 3 dedicated PC's(VMS) with SQL servers 2017 were actually equivalent to the configuration with SQL server 2014 running all DBs (8) on single VM.