I think this questions is already asked by one of the members and have nicely been discussed at:

https://www.researchgate.net/post/How_can_I_calculate_the_load_of_a_virtual_machine/1

Hope the above discussion makes it easier for you.

Good luck

I personally don't like the above discussion because I have found that organizations generally aren't interested in migrating virtual machines (publication eventually on its way on that one).  Instead, most cloud based apps will typically do things like partition applications to stateful and stateless sections, and start up or shut down VMs (usually for handling the stateless work) depending on load.  You can automatically do this kind of scaling on the Amazon Web Services (AWS) cloud and others.  Your question is about what load parameters to use for load work, and I suggest taking a look at what AWS parameters are used for autoscaling.  I have included links to autoscaling, cloudwatch (which has the load parameters for autoscaling) and also a link to design patterns on how people design robust cloud applications.

https://aws.amazon.com/autoscaling/details/

https://aws.amazon.com/cloudwatch/details/#amazon-ec2-monitoring

http://www.opendatacenteralliance.org/docs/architecting_cloud_aware_applications.pdf

Thanks for your valuable answers.