Metrics describe an aspect of a system at a particular point in time and are displayed in numerical values. They are collected at regular intervals and are identified with a timestamp, a name, a value, and one or more defining labels. They are capable of supporting near real-time scenarios and are useful for alerting. Alerts can be fired quickly with relatively simple logic.

Metrics in Azure Monitor are stored in a time-series database that is optimized for analyzing timestamped data. This makes metrics suited for the fast detection of issues. They can help to detect how your service or system is performing, but to get the overall picture, they typically need to be combined with logs to identify the root cause of issues.

You can use metrics for the following scenarios:

• Analyzing: Collected metrics can be analyzed using a chart in Metric Explorer. Metrics from various resources can be compared as well.
• Visualizing: You can create an Azure Monitor workbook to combine multiple datasets into an interactive report. Azure Monitor workbooks can combine text, Azure metrics, analytics queries, and parameters into rich interactive reports.
• Alerting: Metric alert rules can be configured to send out notifications to the user. They can also take automatic action when the metric value crosses a threshold.
• Automating: To increase and decrease resources based on metric values that cross a threshold, autoscaling can be used.
• Exporting: Metrics can be streamed to an Event Hub to route them to external systems. Metrics can also be routed to logs in the Log Analytics workspace, to be analyzed together with the Azure Monitor logs and to store the metric values for more than 93 days.
• Retrieving: Metric values can be retrieved from a command line using PowerShell cmdlets and the CLI, and from custom applications using the Azure Monitoring REST API.
• Archiving: Metric data can be archived in Azure Storage. It can store the performance or health history of your resource for compliance, auditing, or offline reporting purposes.

There are four main sources of metrics that are collected by Azure Monitor. Once they are collected and stored in the Azure Monitor Metric database, they can be evaluated together regardless of their source:

• Platform metrics: These metrics give you visibility of the health and performance of your Azure resources. Without any configuration required, a distinct set of metrics is created for each type of Azure resource. By default, they are collected at a one-minute frequency. However, you can configure them to run on a different frequency as well.
• Guest OS metrics: These metrics are collected from the guest operating system of a virtual machine. To enable guest OS metrics for Windows machines, the Windows Diagnostic Extension agent needs to be installed. For Linux machines, the InfluxData Telegraf Agent needs to be installed.
• Application metrics: These metrics are created by Application Insights. They can help to detect performance issues for your custom applications and track trends in how the application is being used. 
• Custom metrics: These are metrics that you define manually. You can define them in your custom applications that are monitored by Application Insights or you can define custom metrics for an Azure service using the custom metrics API.