Creat/Configure VMWare Clusters.

Question 1

DRS VM Entitlement

Answer 1

When a cluster is overcommitted, DRS and VMKernel work together to allocate resources based on the resource entitlement of each VM. Based on configured shares, configured memory size, reservations, current demands on VMs and Resource Pools that contain them.

Question 2

Configuring VM Component Protection.

Answer 2

1. PDL (Permanent Device Loss): Occurs when storage array issues a SCSI sense code informing that the device is unavailable. Host cease I/O requests to that array.
2. APD (All Paths Down): Host cannot reach storage but no PDL SCSI sense code is received either. Host keeps trying until a timeout is reached. 140 seconds by default. Can be changed using Misc.APDTimeout.

Question 3

PDL Response

Answer 3

1. Disabled: No action will taken, no notification.
2. Issue Events: No action, administrator notified.
3. Power Off and Restart VMs: Attempt restart.

Question 4

APD Response.

Answer 4

1. Disabled: No action taken.
2. Issue Events: No action, administrator notified.
3. Power Off and Restart VMs (conservative): Won’t attempt restart until it finds another host with storage available to restart VMs. Communicates with master host.
4. Power Off and Restart VMs (aggressive):

Question 5

Delay for VM failover for APD.

Answer 5

In addition to the 140 timeout. Default is 3 minutes.

VMCP will not take action for 5 minutes, 20 seconds (Timeout plus delay). VMCP Timeout.

Question 6

Response for APD Recovery After APD:

Answer 6

Action to take after Storage event clears after 140 second timeout but before 3 minute additional delay.

1. Disabled: No action taken. VMs will be able to connect to storage.
2. Reset VMs: Wil cause a hard reset of the VMs, try to reconnect to storage. Short outage during reset.

Question 7

Configuring Migration Thresholds for DRS and VMs.

Answer 7

Figure 14-25.
1. “Satisfy cluster constraints like affinity rules and host maintenance”. Priority 1 recommendations.

2. Priority 1 and 2 recommendations. “Promise a significant improvement to the cluster’s load balance”.
3. Priority 1, 2 and 3 recommendations. “ promise at least good improvement to the cluster’s load balance”.
4. Priority 1, 2, 3, and 4 recommendations. “promise even a moderate improvement to the cluster’s load balance”.
5. “promise even a slight improvement to the cluster’s load balance”.

Question 8

Configuring Automation Levels for DRS and VMs.

Answer 8

Manual: vCenter makes recommendations about initial placement and load balancing.

Question 9

Configuring Automation Levels for DRS and VMs.

Answer 9

Manual: vCenter makes recommendations about initial placement and load balancing.

Partially Automated: vCenter chooses the right host for initial placement of the VM. No recommendations in regard to initial placement but recommendations are made with regard tp load balancing.

Full Automated: vCenter makes al decisions.

DRS settings can apply to cluster but be over-ridden by individual hosts. Figure 14-25,14-26.
Must select “Enable Individual VM Automation Levels”.

Question 10

Configure Admission Control for HA and VMs.

Answer 10

vSphere HA provides for the automatic restart of VMs on another host when the host on which the are running has failed or has become disconnected from the cluster.

Admission Control: Host holds back resources to assist in the event of failure of different host. Amount of resources if configurable.

Question 11

Admission Control.

Answer 11

Enabled by Admission Control Policy
If you need to start a critical VM and system won’t let you, you can use “Do Not Reserve Failover Capacity” option to temporarily override the HA Admission Control.

Question 12

Admission Control Policy.

Answer 12

1. Define failover capacity by static number of hosts: 1-31, usually not set above 2, usually set to 1.
   Uses “slot size”: estimate of CPU and RAM needed for every VM in cluster.
2. Define failover capacity by reserving a percentage of the cluster resources: Set a percentage for org. Percentages can be different for CPU from memory.
   HA continually compares total resource requirements for all VMs with total failover capacity that it ahs derived from your settings.
3. Use dedicated failover hosts: Passive standby hosts. (HA supposed to provide fault tolerance w/o need for standby hardware).

Question 13

VM Overrides

Answer 13

Pg. 589
1. VM Restart Priority: Default is Medium for all VMs so if left at default, all will try to restart.

2. Host Isolation Response: Determines host’s response when host is not receiving heartbeats on management network.
   Options: (Can be set for each VM in cluster).
   Leave it powered on. (Default)
   Power Off.
   Gracefully shutdown VMs.

Question 14

Determine Appropriate Failover Methodology and Required Resources for HA.

Answer 14

Difference between Admission Control methods.
1. Host Failures the Cluster Tolerates:
“Define failover capacity by static number of hosts”.
Uses largest reservations of CPU and memory.
Values used to assume every VM is that size.

2. Percentage of Cluster Resources as Failover Spare Capacity:
   “Define failover capacity by reserving a percentage
   of the cluster resources”.
   Works well for VMs that have highly variable CPU
   and memory usage reservations.
   Can configure separate percentages for CPU and
   memory.
3. Specify Failover Hosts:
   “Use dedicated failover hosts”:
   Some organizational policies dictate passive
   standby hosts. Provided for flexibility, bit best
   practice.