Ben Yemini

Automating Instance Suspension in AWS or Azure

Blog Post created by Ben Yemini Expert on Nov 7, 2017

Turbonomic can automatically suspend unused virtual machines in an AWS or Azure environment. This can significantly reduce monthly bills and eliminate the need to chase down development teams to clean up their environments. 

 

Here is an example of how to enable this capability in your environment with Turbonomic version 6.1. 

 

Navigate to Settings > Groups. 

 

Create a new application group and select the apps to include. See example below. 

Now you need to set an Automation Policy for the application group you just created. This can be found under Settings > Policies and selecting Automation Policy. 

 

Select Application under policy type and define the following:

  • Scope to the app group you defined 'auto suspend group' in my example above  
  • Set the schedule, e.g. daily 5 pm to 11 pm 
  • Set Application Priority to Normal 

 

See example below. 

 

Finally you need to set the Minimum Sustained Utilization and ensure the underlying VMs for these application workloads have the right automation policy in place to stop and start (for proof of concept you may want to first set this to Recommend or Manual before moving to full automation). 

 

To the set the VM automation policy navigate to Settings > Policies and selecting Automation Policy. 

 

Select Virtual Machines under policy type and define the following:

  • Scope to the underlying VM group for the apps 
  • Set the Minimum Sustained Utilization to a current and historical VCPU utilization value below which you'd like the VM to suspend 
  • Set Suspend under Action Automation to Automated, Manual or Recommended 
  • Set Start under Action Automation to Automated, Manual or Recommended 

 

See examples below. 

 

 

That's it. Turbonomic will identify apps that are not consuming resources during the time window you define and drive suspend actions on the underlying VMs. And when the policy time expires Turbonomic will turn those VMs back on. 

Outcomes