3. Organization Virtual Datacenter Examples : 3.2 Reservation Pool Model : 3.2.4 Design Implications
3.2.4 Design Implications
Because this allocation model assigns all settings on a resource pool basis, updates to this model do not require a shutdown and restart of the virtual machines to pick up the changes. Based on the settings chosen, this model could be considered as one of the best models because the consumer has control over how the pool of resources is divided among the virtual machines.
The provider must monitor the use of the Maximum number of VMs setting. If this is left as unlimited and the consumer does not set any per virtual machine reservations, there is potential for overcommitment. A consumer can, in theory, deploy more or larger virtual machines than the pool is configured for (additional resources such as memory could be paged out to satisfy the configuration). Assigning a limit to the maximum total number of virtual machines can help to keep the pool within the expected size.
Huge virtual machines can be created using this model, but it is important to understand the role that overhead reservations play along with the standard per virtual machine reservations. Understanding the effect of these settings on vSphere is crucial for implementation.
For more information, see Chapter 13 in the book VMware vSphere 5.1 Clustering Deepdive, available from retailers.