The key part of the article for your question is
"To give you an example, this is what the iterations could look like as we live-migrate a virtual machine with 24GB of memory:"
they don't say a VM with 24gb of memory actively being used, they say a VM with 24GB of memory. I'm not sure the purpose of this question other than understanding how VM works, are you wondering if in the end vmotion leaves memory on the older host? It doesn't, post vmotion it still requires the same resources as it did before. One thing I think will help if you consider things like drs
vSphere 6.5 DRS and Memory Balancing in Non-Overcommitted Clusters - frankdenneman.nl
scroll down to the "Out-of-the-box DRS Behavior" setting. IF you leave the default drs setting in 6.5 and 6.7(I think 7.0 it changed), drs only considers active plus an extra 25% or the idle consumed memory when making a decision to move a VM or not. The "Memory Metric for Load Balancing Enabled" on the other hand uses the full consumed memory when deciding to move a vm.
This is important because with the default setting a VM with large amount of VM provisioned, but is mostly inactive , could be migrated to a host and you could have more memory provided then you actually have. This can be fine in some situations, but if that VM suddenly wants all that memory you can start to have performance issues., If you have the "Memory Metric for Load Balancing Enabled" but you can run into issue here if you overcommit your memory as vms won't consolidate like they did before.
