Azure VM metrics collection in batch mode

Turbonomic can collect memory, CPU, and other metrics for a large number of VMs when you enable VM metric collection in batch mode.

VM metric collection in batch mode is optional and is disabled by default. When enabled, Turbonomic makes a single API call to collect metrics for 50 VMs (this number is configurable), instead of a single API call for each VM. This can reduce the likelihood of Azure throttling API requests from Turbonomic, which happens when the number of API calls exceeds a certain limit. Throttling can sometimes result in missing metrics in Turbonomic.

To enable VM metric collection in batch mode:

1. Enable the following property for the mediation-azure probe.

   metrics.batch.api.enabled

   In the Turbonomic Swagger API, set the property to true on the probe of type Azure Subscription.

   Contact your Turbonomic representative if you need assistance setting the probe property.

2. Assign the Monitoring Reader role to the subscription containing the VMs for which metrics will be collected in batches.

   Azure requires this role to enable the collection of metrics through the Azure Batch APIs. This role can view all monitoring data in a subscription but cannot modify any resources or edit any settings related to monitoring resources. For details about roles, see the Azure documentation.