Rightsizing the Instances in your Runbook

For any Run in Cloud or Migration operations, you must define the target instance type in your runbook inventory file. You can do this manually, but you can also use Velostrata's Rightsizing capability to accomplish this. Using this capability only works for rows whose RunGroup value is set to a positive number. 

Rightsizing can provide two types of recomendations for you, as follows:

  1. Performance-based: this will create a suggestion based on the CPU and RAM options currently selected for the on-premises VM. You can get this recommendation without having previously toggled on rightsizing monitoring. 
  2. Cost-based: this will evaluate both the current CPU and RAM options of the on-premises VM alongside the average usage of this VM over a period of time. To use this option, you must have already toggled on 'rightsizing monitoring' for this group of VMs. It is recommended that monitoring run for at least 48 hours to provide accurate insight, though this could be shorter or longer depending on usage.

Your rightsizing options will automatically be added to your inventory file via a number of new columns that are added. There will be six recommendations made in total. 3 that reflect performance-based options and 3 that reflect cost-based options (assuming you've toggled on monitoring already). Each recommendation will include two columns: one for the recommended instance target and one that estimates the hourly cost to run that VM. There will also be another column that contains the information on when monitoring began. In total, there will be 13 new columns added to your runbook. The naming convention of these new columns will be as follows

  • PerfOptRec_type_N (N = 1, 2, 3) – the recommended performance-optimized’ cloud instance type.
  • PerfOptRec_cost_N (N = 1, 2, 3) – the list price for the recommended cloud instance type.
  • VmUsageInfo – the collected VM usage information – this is the base for cost-optimized recommendations.
  • CostOptRec_type_N (N = 1, 2, 3) – the recommended cost-optimized cloud instance type.
  • CostOptRec_cost_N (N = 1, 2, 3) – the list price for the recommended cloud instance type.

Here is an example of what you'll see in your inventory file after running rightsizing:




















VmUsageInfo:[MonitoringStartDate: 11/4/2017 6:46:59 AM, HistoricalDayCount:0.00347222222222222, CpuUsage: 10, MemoryUsageGB: 1.75]







Here is how to perform the appropriate rightsizing actions:

1. From within the Runbook Automation main menu, click the Rightsizing button.

2. Click the Choose File button, navigate to your inventory CSV file, and select it.

3. Using the Operation drop-down menu, select the operation you'd like to perform:

  • Start Monitoring: this will begin to monitor the on-premises VMs in this runbook inventory file so that you can receieve cost-based recommendations at a later date.
  • Stop Monitoring: this will turn monitoring off for the on-premises VMs in this runbook inventory file once you've successfully retrieved your recommendations. The maximum number of VMs that can be monitored at one time is 100, so it is important to turn monitoring off once recommendations are retrieved and runbook jobs are complete.
  • Get Recommendations: this will take your runbook inventory file and populate the 12 recommendation columns (as detailed above).

4. Once you've successfully performed 'Start Monitoring' and 'Get Recommendations', you can proceed to making your selections by opening the runbook inventory file in your preferred spreadsheet editor. 

5. Locate the recommendation columns (listed in the table in the middle of this article) and examine their suggestions. Pick the one that you would like to use for any given row and copy the instance type value. 

6. Paste that instance type value (from step 5) into the column for VM that is titled 'TargetInstanceType'.

7. You do not need to delete the recommendation columns before executing your runbook job. The runbook will ignore them.

8. Proceed to use this modified runbook inventory file to run jobs (see next sections).