Which virtual machine (VM) sizing strategy will provide the highest level of uptime, individual VM performance, and failover capacity?

A developer is tasked with building an application to process shipping requests. The developer is consulting the vSphere team to determine failover options and performance best practices.
The development team is providing three physical ESXi hosts with 8 CPU cores and 256GB of RAM per host. The developer does NOT know how many virtual machines they will require.
Which virtual machine (VM) sizing strategy will provide the highest level of uptime, individual VM performance, and failover capacity?
A. A few large 8 vCPU VMs per host protected by vSphere HA.
B. Many small 1 vCPU VMs participating in an OS level clustered application protected by vSphere HA.
C. A few large 8 vCPU VMs per host protected by vSphere Fault Tolerance.
D. Many small 1 vCPU VMs participating in an OS level clustered application protected by vSphere Fault Tolerance.

microsoft-exams

9 thoughts on “Which virtual machine (VM) sizing strategy will provide the highest level of uptime, individual VM performance, and failover capacity?

  1. A) Matching vCPU with the physical CPU cores provide best performance. However, we don’t know how many VMs are required, therefore there’s a risk of a high CPU overcommitment (especially in case of failover), which will impact on performance. Also, if the application doesn’t need/use 8 cores, their use will be scheduled anyway therefore wasting resources (we don’t know that, but if 1 vCPU is enough for the application then it would be counterproductive to configure it with 8)
    B) It’d be good practice to start a VM with lower vCPU possible and then increase if needed. This solution lower the risk of over-commitment, and guest clustering is supported for HA. Might or might have lower performance, requirements of the application are not specified
    C) No, FT supports up to 4 vCPU
    d) No, FT doesn’t support OS (Microsoft at least) failover clustering

  2. Going with A too.

    C is not correct as stated by others. FT doesn’t support 8vCPU
    D is not correct as FT is not supported with OS level cluster (MSCS at least)

    B does not provide ‘individual’ vm performance but rather an army of small VM’s

Leave a Reply

Your email address will not be published. Required fields are marked *


The reCAPTCHA verification period has expired. Please reload the page.