Hi All 
...
In the twelfth part of our series, we'll go through first main Clustering feature in vSphere: Distributed Resource Scheduler (DRS) and its supplementary feature: Distributed Power Management (DPM).
Credits:
- Frank Denneman
Now, Let's Start...
1. DRS Resources Distribution Chart:
Resources Distribution Chart is a nice chart that shows information about consumption of each VM inside DRS-enabled Cluster of resources (CPU/Memory). Unfortunately, it’s somehow hard to get these information out of it. The following article by Frank Denneman describes in details this chart and how to get what you need out of it:
http://frankdenneman.nl/2010/03/08/drs-resource-distribution-chart/
Although, it’s written about vSphere 4.0, but it applies also on vSphere 5.0 and later.
2. DRS Recommendations and Actions:
DRS recommendations or actions are initiated by satisfying the next two conditions together:
1-) DRS Entitlement of any resource of a VM isn’t met.
2-) Cluster load across all of the hosts is imbalanced.
Put in mind that, if a cluster can give all VMs their resources entitlements sufficiently will not initiate a DRS recommendation or action even if the cluster is load imbalanced. In some corner cases also –when using Reservations, Shares& Limits excessively-, a VM can’t get all its resources entitlements although the cluster is balanced. In this case, DRS recommendations or actions are initiated.
3. DRS vs. Manual Power Commands:
Manual reboot or shutdown will not initiate any DRS actions or recommendations to migrate any VM hosted on it. To shut down or reboot a host manually, you have to enter it Maintenance Mode which will initiate DRS actions or recommendations to migrate any VM hosted on that host.
4. Affinity Rules:
Affinity Rules are used to add some restrictions on DRS behavior and migrations of VMs. It can be either to keep some VMs together (VM-VM Affinity Rule), separate some VMs while migrations (VM-VM Anti-Affinity Rule) or keep a group of VMs on certain group of hosts (VMs-Hosts Affinity Rule).
Affinity Rules have two restriction levels:
1-) Should: DRS Cluster tries to fulfill the rule, but if it can’t, it’ll override the rule in case of host failure or no resources available.
2-) Must: DRS Cluster must do the rule or the VM won’t be powered on.
Keep in mind that, VM-VM affinity rules shouldn’t conflict with each other. Meaning, you shouldn’t have one rule that separates virtual machines and another rule that keeps them together. If you have conflicting rules then the older rule wins and the new rule is disabled.
Another thing, DRS gives higher precedence to prevent violations of anti-affinity rules than violations of affinity rules. It means that if DRS engine has to violate a rule, it’ll violate affinity rules first and if it doesn’t fulfill its needs, it’ll violate anti-affinity rules as a last resort.
Last, manual reboot, shutdown or entering Maintenance Mode won’t violate Must (Not) affinity Rules.
5. Distributed Power Management (DPM):
DPM is a supplementary feature with DRS to save power consumption of ESXi hosts in a DRS-enabled cluster in case of low loads. It shuts down hosts and doesn’t put them in stand by (sleep) mode as shown in vSphere (Web) Console. When load increases, it begins to power them up one by one according to the need.
DPM waits 40 mins before shutting down any host to make sure that the load is really low before shutting down any host in order to avoid powering up and down hosts rapidly.
DPM doesn’t always shut down the same host every time, but it checks the attribute (Last Time Exited Stand By - DPM Hosts Settings) and powers down the host with the longest time up.
With Wake-on-Lan (WoL) NICs used to power on Stand-by Hosts, WoL Magic packet is sent to ESXi hosts on VMKernel ports with vMotion enabled only.
The following technical paper from VMware explains DPM feature deeply enough:
http://www.vmware.com/files/pdf/Distributed-Power-Management-vSphere.pdf
Share the Knowledge ...
Previous: vSphere 5.x Notes & Tips - Part XI: