Azure VM memory metrics collection

It is highly recommended that you enable collection of metrics in your environment. Enabling metrics allows Turbonomic to generate scale actions to optimize VM resource usage. For Turbonomic to collect metrics, you must enable the collection of these metrics on the VMs in your environment.

Turbonomic collects memory metrics for Azure VMs using the following mechanisms, in order:

Log Analytics workspace - InsightsMetrics table

This is the preferred collection method. This mechanism is used when a VM is configured with the Azure Monitor Agent (AMA). This agent collects guest OS memory metrics and sends it to the Log Analytics workspace InsightsMetrics table. Turbonomic then collects available memory metrics by querying this table.

For more information on the Azure Monitor Agent, see the Azure documentation.
Log Analytics workspace - Perf table

This mechanism is used when a VM is configured with the Log Analytics agent (also known as MMA and OMS) and is connected to a Log Analytics workspace. This agent sends guest OS memory metrics to the Log Analytics workspace Perf table. Turbonomic then collects available memory metrics by querying this table.

Note:
Azure retired the Log Analytics agent. For information on the recommended migration to the AMA, see the Azure documentation.

If your Log Analytics workspaces are not in the same subscription as VMs, contact your Turbonomic representative for assistance.
Diagnostics monitoring based on the storage account SDK

This legacy mechanism installs an Azure diagnostics extension on the VM, which sends guest OS level memory metrics to a designated storage account table. During the discovery process, Turbonomic collects VM memory from these tables. For this mechanism, Turbonomic needs the Storage Account Contributor role to access the storage account keys and establish a connection that can retrieve VM memory statistics.

For information about enabling this memory mechanism in the Azure portal, see this IBM support page.
Metrics REST API

If memory cannot be collected using the previously-mentioned mechanisms, Turbonomic tries to collect host OS based memory for the VM through REST API calls, which is the same mechanism used for collecting other VM metrics, such as CPU usage. Turbonomic collects VM memory and generates memory-driven scale actions where appropriate.

Note:
Azure provides available memory through the REST API (in preview mode) or the Azure portal (in a VM’s Monitoring > Metrics section).

Memory source groups for Azure VMs in Turbonomic

In Turbonomic, memory source groups reflect the source of Azure VM memory metrics. To search for a specific memory source group in the Turbonomic user interface, open the Search page, choose Groups, and enter memory in the search field. Each group contains the list of VMs for which memory was collected through a specific mechanism. These are dynamic groups that are continuously updated based on each discovery.

The Azure Memory Monitoring Unavailable group contains VMs for which memory could not be collected using any of the collection mechanisms. This group may contain VMs that are not in a RUNNING state. For these VMs, metrics data is often not available and memory-driven scale actions are not generated. However, other scale actions may be generated.