Two sibling resource pools named RP-CHILD-A and RP-CHILD-B have these configurations: - RP-CHILD-A has an expandable reservation and memory shares set to 2000' with 10 VMs, each with 4GB of RAM…

The pools have different absolute share values (2000 vs. 4000), so they do not receive the same entitlement - RP-CHILD-B receives twice the pool-level allocation of RP-CHILD-A.

vSphere always uses shares to arbitrate resource allocation during contention; the expandable reservation setting only allows borrowing unused capacity from a parent pool, not bypassing share-based arbitration.

RP-CHILD-A has 2000 shares divided among 10 VMs, yielding 200 effective shares per VM, while RP-CHILD-B has 4000 shares divided among 30 VMs, yielding roughly 133 effective shares per VM. As long as no VMs are added or removed, each VM in RP-CHILD-A will consistently receive proportionally more memory than each VM in RP-CHILD-B during contention because of this per-VM share advantage.

VMs in RP-CHILD-B receive fewer effective shares per VM (approximately 133) compared to RP-CHILD-A (200) because the larger share pool is split across three times as many VMs.