Database Reference
In-Depth Information
Figure 10.3
Example of compatible blended database peak workloads.
vMotion Slot Sizes of Monster Database Virtual Machines
vMotion Slot Size is a term coined to describe how much compute capacity is required
in order for the largest VM to migrate from one host to another in a VMware DRS
cluster or for the largest VM to restart in the case of a VMware HA event. The vMotion
Slot Size should not be confused with a VMware HA Slot Size because it needs to be
considered regardless of whether your VMware HA Admission Control Policy is set to
Number of Host Failures or Percentage of Resources Reserved for Failure. It is
especially applicable to large database servers that will likely have large memory
reservations.
If there is enough overall compute capacity in your cluster for maintenance or HA
events but this capacity is fragmented across the hosts, it may impact your ability to put
hosts into maintenance mode. It may also extend the time it takes for HA to restart your
largest VMs. In the case of an HA event, DRS will kick in and attempt to defragment the
cluster's resources in order for the VMs to be able to restart; however, there is no
guarantee it will be able to free up enough resources.
In order to ensure you always have enough free resources in your clusters to perform
maintenance and restart in the case of an HA event, it is recommended as a general rule
of thumb that your ESXi hosts are twice the size of the largest VM you plan to support.
