>>Are the virtual servers all running on the same physical server? Are they provisioned with the same amount of RAM and number of virtual CPUs?
>
>Yes and Yes
>
>>Can you monitor Task Manager and view performance/resource use while the service loads on different servers?
>
>In Performance\CPU Usage, I can see that the servers who perform well will use more CPU to load the Windows Services. I did the test with four Windows Services on each server. So, this extra CPU allowed seems to give a boost of performance. For them, the CPU gets between 12% and 16% everytime I launch the command. For the servers not performing well, the CPU allowed ranges from 6% to 7%. This seems to reflect the difference I have observed. Does this tell you something interesting? :)

Hmm, sounds like you may have to dive deeply into how the specific VMs for the servers are configured. I believe some hypervisors are able to set higher priorities for certain VMs regardless of whether all VMs are provisioned with the same nominal RAM and virtual CPUs. This can be important in hypervisors that allow over-provisioning i.e. sum total of provisioned CPUs and RAM is higher than the actual server hardware.

It may be that the "slow" servers are somehow set to a lower priority than the "fast" ones.

You may also see this concept referred to as hypervisor quality of service (QoS).

>>Are there any Event Log messages during service start, such as calls failing or timing out due to improper or incorrect configuration? e.g. configured with different primary and secondary DNS servers, configured to default to IPv6 vs. IPv4 etc. You don't necessarily need to know how these things affect performance, just look for configuration differences.
>
>No, there is no event log when I start the Windows Services other than the fact that it logs a stop or a start.