Idle VM recommendation

The idle VM recommendation identifies VMs that are idle or unused. A VM is detected to be idle if both the following criteria are met:

• The VM exists for at least 7 days.
• The VM is powered off or the CPU utilization, network rate, and disk rate of the VM are less than their corresponding thresholds for at least 95% of the analysis period.

The idle VM recommendation is available for these technologies: AWS, Azure, and GCP

The following recommendation details are provided:

• Recommended action that suggests removing the VM to achieve the following benefits:
  
  • Reclaim storage
  • Reduce costs
  • (For AWS only) Reclaim a slot in the free tier if the VM is within a free tier
  • (For AWS only) Reclaim a reserved slot if the VM is reserved
• Duration and time for which the VM was idle
• Duration for which the VM was powered on and powered off
• Analysis period
• Criteria for the recommendation

You can analyze these recommendation details and decide the action to be taken on the VM. For example, you can decide to remove the VM in the following scenarios:

• The VM is allocated to the project that was already completed.
• The VM is allocated to the upcoming project that is getting delayed.

In some scenarios, you might decide to wait before removing the VM. For example, if the VM is running in a production environment, you can verify whether other resources can be allocated to the VM instead of removing it. 

The recommendation rule is based on the following indicator thresholds. 

• CPU utilization = 5 %
• Disk I/O rate = 100 Kbytes/s
• Network rate = 200 Kbytes/s

You can change the default values of these thresholds on the Thresholds page in the Helix Capacity Optimization Console. 

For more information about configuring thresholds, see Configuring and managing thresholds for metrics and indicators.

The Optimizer rule identifies a VM as idle if the values of these indicator thresholds are less than the 95^th percent of the time during the analysis period. If you want the Optimizer rule to detect the VM as idle during the entire analysis period, you can change the idle VM powered on time value to 100 percent on the Rules page in the Helix Capacity Optimization Console.

For more information about modifying the Optimizer rules, see Configuring and managing Optimizer rules.