Provision New Physical Machine – Is it because of HA Admission Control?

Document created by Anson McCook Expert on Jul 3, 2015Last modified by Anson McCook Expert on Nov 1, 2016
Version 3Show Document
  • View in full screen mode

Are you seeing an action to “Provision New Physical Machine” in some of your clusters? If utilization levels of resources seems “normal” but you still see this recommendation it could be caused by your HA Admission Control settings.

 

Turbonomic imports your HA Admission Control settings to ensure you are always in compliance of your business policies. To see if HA is causing the Provision action, check out the Policy Tab.

 

1) Policy Tab > Analysis > Hosts > Folders

 

Notice: If you see an asterisks (*) next to the “Folders” drop-down, and proceeding Host groups, that is an indicator that at least one cluster in the environment has Admission Control configured.

 

2) Drop down into a host group and take notice of any bold clusters. Bold indicates that a cluster has Admission Control configured.

3) Under “Utilization Constraints” you’ll notice the values for Host CPU Utilization and Host Memory Utilization will be modified – as in, will not be the default 100% value.

 

HA1.png

 

Think of the Utilization Constraints as a “cliff.” It’s the place you never want to reach. The “Desired State” slider defines “how far away” you want to be from your cliff. With Admission Control enabled, you are reserving a certain amount of resources so that you can support a fail-over of any number of hosts. Essentially, the distance to your cliff has shrunk because resources need to be reserved. By incorporating this reservation into Turbonomic’s actions, it’s able to let you know when capacity needs to be added to continue to support your fail-over settings.

 

So if you check out host utilization levels (perhaps on the Optimize Dashboard or Inventory Tab) and look at utilization levels for Memory and CPU, you’ll notice Turbonomic is recommending to Provision New Physical Machines if those hosts are close to or beyond the Utilization Constraints you observed in the policy tab.

 

Here’s a quick example to show how Turbonomic calculated the new Utilization Constraint for Memory and CPU.

 

I have a 4 host cluster with my Admission Control Policy set to tolerate 1 host failure.

 

This means the reservation across 4 hosts will be set at 25%. Turbonomic will now make sure to set the Utilization Constraint at 75% for the cluster. The equation is simply:

X/N = Reservation %

Where X is the number of "Host Failures the Cluster Tolerates"

Where N is the number of hosts in the cluster.

 

And it follows that:

1-(X/N) = Utilization Constraint %

 

HA2.png

 

HA3.png

1 person found this helpful

Attachments

    Outcomes