What Is Azure Monitoring? How Azure Monitor and Alerts Work
Most Azure problems are not caught by Azure. They are caught by a customer, a colleague or a Monday morning phone call. A virtual machine runs out of disk over the weekend, an application slows down after an update, a storage account starts failing requests, and nobody knows until someone complains. Azure monitoring exists to close that gap.
Azure Monitor is Microsoft’s built-in monitoring service for Azure, hybrid and multicloud environments. It collects metrics, logs and events from your resources, helps you analyze them, and sends alerts when something needs attention. Azure monitoring is the practice of using it well: knowing what to watch, which alerts matter and who responds.
This guide explains what Azure Monitor does, how the data flows, what it collects from virtual machines (and what it quietly does not), the alert types, how to set up your first alert, what it costs and where managed support fits. If you are new to the platform itself, start with our primer on what Azure is and how it works.
Key Takeaways
- Azure Monitor is the platform, monitoring is the practice. The service collects and analyzes data. Someone still decides what to watch and responds.
- Some data arrives automatically, some does not. Host CPU, disk operations and the activity log need no setup. Memory and disk space inside a VM need the Azure Monitor Agent.
- Alerts come in different types. Metric alerts watch numbers, log alerts watch query results, activity log alerts watch changes, and Service Health alerts watch Azure.
- An alert is only a signal. Action groups notify or automate, but triage, diagnosis and the fix still need a person or a good runbook.
- Logs drive the cost, alerts do not. Platform metrics and the activity log are free. Log ingestion is billed per GB after the first 5 GB, while an alert rule costs a few dollars a month at most.
- Too many alerts is as risky as too few. Alert fatigue hides the real incident. Fewer alerts, each with an owner, work better.
In this guide:
- What is Azure monitoring?
- What is Azure Monitor?
- How Azure Monitor works
- What Azure Monitor collects from a virtual machine
- Tools, insights and how they differ
- What is an Azure Monitor alert?
- Severity, states and alert behavior
- Action groups and alert processing rules
- Azure Service Health: is it Azure or is it you?
- How to set up your first alert
- How to avoid alert fatigue
- What happens after an alert fires
- How much Azure Monitor costs
- Common monitoring mistakes
- Do it yourself or get managed monitoring?
- How Apps4Rent helps
- Frequently asked questions
What Is Azure Monitoring?
Azure monitoring means continuously observing your Azure environment to confirm that resources and applications are working as expected. It answers four plain questions: is it up, is it fast enough, is it running out of room, and did something change? Depending on what you run, monitoring can cover:
- Virtual machines, including CPU and memory usage
- Network activity and connectivity
- Storage usage and availability
- Database performance
- Application and system logs
- Azure service health and planned maintenance
The objective is not to collect data for its own sake. It is to turn data into information that helps IT teams spot problems early, fix them quickly and keep workloads reliable. A company with two virtual machines can check them by hand. Add applications, databases, storage and users, and it becomes hard to know what is happening without a system doing the watching.
What Is Azure Monitor?
Azure Monitor is the Microsoft service that does the collecting, storing, analyzing and alerting. Think of it as a central monitoring layer: telemetry from across the environment lands in one place, and you work from there. It supports virtual machines, applications, containers, Azure services and hybrid servers, and it helps answer three questions:
- What is happening? Current and historical performance and health.
- Why is it happening? Logs, traces and correlated events that point to a cause.
- What should we do about it? Alerts, notifications and automated responses.
Take an accounting application running on an Azure virtual machine that suddenly slows down. Azure Monitor can show whether the cause is CPU saturation, low memory, a full disk or a failing dependency. That visibility lets the IT team fix the problem before users start phoning in.
Azure Monitor works with four main kinds of data:
| Data type | What it is | Example | Where it lives |
|---|---|---|---|
| Metrics | Numeric values sampled at regular intervals | CPU percentage every minute | Azure Monitor metrics database |
| Logs | Timestamped records with detail | A Windows event, a failed sign-in, an application error | Log Analytics workspace |
| Traces | Records that follow a request through an application | A slow web request and the database call behind it | Application Insights |
| Changes | Records of what changed and when | A configuration update or a deleted resource | Activity log and change tracking |
One point surprises many people: Azure Monitor is not something you buy or install first. It is on for every subscription. Platform metrics and the activity log are collected automatically at no charge. What you add is the extra data, such as guest operating system counters and application logs, plus the alert rules that turn data into action.
How Azure Monitor Works
Data moves through five stages, and knowing them makes every setting easier to place.
- Sources. Your virtual machines, applications, databases, storage accounts, networks, containers and subscription events produce telemetry.
- Collect. Platform metrics flow in automatically. Everything else needs a mechanism: the Azure Monitor Agent for data inside a virtual machine, diagnostic settings for resource logs, or instrumentation for applications.
- Store. Metrics go into a time-series database. Logs go into a Log Analytics workspace, which you size and retain according to need.
- Analyze. Metrics Explorer charts numbers, Log Analytics queries records with Kusto Query Language (KQL), and workbooks and insights present both in ready-made views.
- Respond. Alert rules watch the data, action groups carry out the response, and automation can fix routine problems without waiting for a person.
Metrics and logs are the two data types you will use most, and they behave differently:
| Metrics | Logs | |
|---|---|---|
| What it holds | Numbers over time | Detailed records and events |
| Best for | Spotting trends and thresholds quickly | Finding the cause of a problem |
| How you query it | Metrics Explorer | KQL in Log Analytics |
| How long it is kept | 93 days for platform metrics | 31 days included, extendable on a pay-as-you-go workspace |
| Cost | Platform metrics are free | Billed per GB ingested after the first 5 GB a month |
What Azure Monitor Collects From a Virtual Machine
Virtual machines are where most businesses start, and where one of the most common monitoring gaps hides. Some data is available the moment a VM exists. Other data is not collected until you ask for it.
| Data | How it arrives | Agent needed? |
|---|---|---|
| Host CPU, disk operations and network | Platform metrics, collected automatically | No |
| Activity log (starts, stops, deletes, role changes) | Collected automatically for the subscription | No |
| Boot diagnostics | On by default for VMs created in the portal | No |
| Guest memory and logical disk space | Measured inside the operating system | Yes: Azure Monitor Agent and a data collection rule |
| Windows events, Syslog, performance counters, IIS logs | Collected from the guest through a data collection rule | Yes: Azure Monitor Agent |
Here is the trap. The portal shows CPU for a new VM straight away, so monitoring looks switched on. Memory and free disk space, which cause a large share of real outages, are not there until the Azure Monitor Agent is installed and a data collection rule tells it what to send. Until then, an alert on low memory simply has no data to evaluate.
VM insights solves this by installing the agent and creating a default data collection rule that gathers a standard set of performance counters. The older Log Analytics agent was retired in 2024, so tutorials that tell you to install it are out of date.
The cleanest fix is to make agent deployment and a baseline alert set part of every new subscription, an idea at the core of a landing zone that starts every subscription with monitoring already in place.
Azure Monitor Tools, Insights and How They Differ
Azure Monitor includes several tools. Knowing which one answers which question saves a lot of searching.
- Metrics Explorer charts metrics from any resource and lets you compare and overlay them.
- Log Analytics is where you write KQL queries against the data in a workspace.
- VM insights adds performance views and optional dependency mapping for virtual machines.
- Application Insights monitors the performance and failures of web applications, including request traces.
- Workbooks and dashboards combine metrics, logs and text into shared reports, and Azure Managed Grafana is available for teams that prefer Grafana.
If your users sign in to desktops hosted in Azure, Azure Virtual Desktop Insights is the monitoring view built on Azure Monitor for session hosts, connection health and user experience.
Azure Monitor is often confused with neighboring services. This table separates them:
| Service | The question it answers | Use it for |
|---|---|---|
| Azure Monitor | Is my workload healthy and performing? | Metrics, logs and alerts across resources |
| Log Analytics | What exactly happened? | Querying log data with KQL |
| Application Insights | Is my application fast and error-free? | Application performance and failures |
| Azure Service Health | Is Azure itself having a problem? | Outages, maintenance and advisories |
| Microsoft Defender for Cloud | Is my environment secure? | Security posture and threat protection |
| Azure Advisor | How could this be improved? | Recommendations on cost, reliability and performance |
Defender for Cloud and Microsoft Sentinel handle security signals, so they sit beside Azure Monitor. Our guide to security practices worth applying in Azure shows how they fit.
What Is an Azure Monitor Alert?
An Azure Monitor alert is a notification or automatic response that fires when monitoring data meets a condition you defined in an alert rule. The rule watches a resource and its data, checks your criteria, and when they are met it fires an alert that can start an action.
Monitoring data → condition met → alert fires → team is notified → someone investigates → issue is fixed
Every alert rule has three core parts: the scope (what is watched), the condition (what counts as a problem) and the action group (who is told and what runs). You also give it a name, a description and a severity.
If a monitored resource shows abnormal usage for fifteen minutes, an alert tells the IT team, who can check whether the workload needs tuning or more capacity. An alert is a signal that a response may be necessary. It is not the response.
Azure Monitor supports several alert rule types. Choose by the kind of data you are watching:
| Alert type | What it watches | Best used for |
|---|---|---|
| Metric alert | Metric values over time: platform, custom and Application Insights metrics | Thresholds on data that is already a metric, such as CPU or availability |
| Log search alert | The results of a KQL query, run on a schedule | Specific events, error patterns and logic that needs a query |
| Simple log search alert | Each log row individually, with no aggregation | Near real-time alerts where every occurrence matters, such as a failed backup job |
| Activity log alert | New events in the activity log | Knowing when a resource is deleted, a role is assigned or a setting changes |
| Service Health alert | Azure outages, maintenance and advisories for your services and regions | Telling a platform problem from one of your own |
Metric and log search alerts can use either a static threshold, a fixed number you pick, or a dynamic threshold, where machine learning learns the normal pattern and alerts on deviations. Dynamic thresholds suit workloads with daily cycles, such as an office system busy at nine and idle at midnight, where a fixed number is either too sensitive at peak or too lax at night.
Severity, States and Alert Behavior
Each alert carries a severity from 0 to 4, set when you create the rule. Use it to decide who is woken up and who simply sees the alert in the morning:
| Severity | Name | Typical response |
|---|---|---|
| Sev 0 | Critical | Page someone immediately: outage or imminent data loss |
| Sev 1 | Error | Respond quickly: a service is failing or degraded |
| Sev 2 | Warning | Open a ticket and investigate during working hours |
| Sev 3 | Informational | Review on a dashboard or in a daily summary |
| Sev 4 | Verbose | Keep for context, rarely worth a notification |
Alerts also track two separate things. The alert condition is set by Azure Monitor: fired when the problem is detected and resolved when it clears. The user response is set by your team: New, Acknowledged or Closed. The two are independent, so an alert can be resolved by Azure yet still sit as New because nobody looked at it.
Behavior can also be stateful or stateless. A stateful alert fires once and stays quiet until the condition resolves, which stops the same problem sending a notification every few minutes. A stateless alert fires every time the condition is met. Metric alerts are stateful by default, log search alerts can be set either way, and activity log alerts are always stateless. Fired alerts are kept for 30 days, so export them if you need a longer history.
Action Groups and Alert Processing Rules
An action group is a reusable list of who gets notified and what runs when an alert fires. Build it once, and attach it to many alert rules.
- Notifications: email, SMS, push and voice call.
- Automation and integration: Automation runbooks, Azure Functions, Logic Apps, ITSM tools, webhooks and Event Hubs.
Automation is where monitoring starts to save real time. A runbook can restart a stopped service, a Logic App can post to a team channel and open a ticket, and a webhook can pass the alert to the tool your support team already uses. Be careful, though: a runbook that restarts a service every five minutes can hide a real fault.
Two habits prevent most notification failures. Send alerts to a shared mailbox or group, never to one person who may be on leave or may have left. And test the action group after you create it, because a notification path you have never tested is one you do not know works.
Alert processing rules change what happens after an alert fires without editing each rule. They can add or suppress action groups, filter by scope or severity, and run on a schedule. The classic use is suppressing notifications during a planned maintenance window, so patching night does not generate a hundred alerts.
Azure Service Health: Is It Azure or Is It You?
When an application stops working, the first question is whether the problem is yours or Microsoft’s. Three views answer it:
| View | Scope | What it tells you |
|---|---|---|
| Azure status | Global and public | Broad Azure-wide incidents |
| Service Health | Personalized to your subscriptions, services and regions | Service issues, planned maintenance, health advisories and security advisories that affect you |
| Resource Health | One specific resource | Whether that resource is healthy, and whether Azure or your configuration is the cause |
Service Health is the one most small businesses never configure, and it is worth ten minutes. A Service Health alert is an activity log alert: you choose the subscriptions, services, regions and event types you care about, then attach an action group. Your team then hears about a regional problem or planned maintenance before a user hits a broken app.
Service Health tells you about the problem, not the recovery. If a regional outage would hurt, the plan for recovering after an outage still has to exist, and our comparison of how backup and Site Recovery divide the work shows where each piece fits.
How to Set Up Your First Azure Monitor Alert
Creating a basic alert takes a few minutes in the Azure portal. This walkthrough uses a virtual machine CPU alert, but the same steps apply to most alert types.
-
Open Alerts in Azure Monitor
In the Azure portal, open Monitor, select Alerts, then choose Create and Alert rule.
-
Choose the scope
Pick what to watch: a single resource, a resource group or a whole subscription. Wider scopes cover new resources automatically but need sensible thresholds.
-
Define the signal and condition
Select a signal such as Percentage CPU. Set the logic, for example average above 85 percent over 15 minutes, checked every 5 minutes.
-
Attach an action group
Select an existing action group or create one with the email addresses, SMS numbers or automation you want to run.
-
Set severity and details
Give the rule a clear name that says what it detects, add a description with the first troubleshooting step, and choose a severity.
-
Review and create
Create the rule. It starts evaluating straight away.
-
Test it
Send a test notification from the action group, and if possible trigger the condition on a non-production resource.
Rules can also be created with Bicep or ARM templates, the Azure CLI or PowerShell, which is how larger environments keep alerts consistent. If you are scripting this, our introduction to scripting routine Azure tasks is a good starting point.
If you are not sure where to begin, start with a small set of alerts that catch the most common failures. Treat the numbers below as starting points to tune, not rules:
| Alert | Type | Starting point | Why it matters |
|---|---|---|---|
| VM availability | Metric | VM stops running | Catches stopped or crashed servers; Azure offers a recommended rule for it |
| High CPU | Metric | Above 85% for 15 minutes | Sustained load slows users; a brief spike does not |
| Low memory | Metric (guest data) | Below 10 to 15% available | Needs the Azure Monitor Agent; a leading cause of slowdowns |
| Low disk space | Metric (guest data) | Below 15% free | A full disk stops applications and databases |
| Backup failure | Simple log search | Any failed backup job | A silent failed backup is found only during a restore |
| Service Health | Activity log | Issues and maintenance in your regions | Separates Azure problems from your own |
| Risky changes | Activity log | Deletes, role assignments, firewall changes | Shows who changed what, and when |
The backup row deserves attention. Monitoring a backup job is cheap, and finding out a backup has been failing for a month is not. Our guide to what Azure backup really costs covers the protection side.
How to Avoid Alert Fatigue
It is tempting to alert on everything. Within a month the team stops reading notifications, and the one that signals a real outage looks like all the others. These habits keep it under control:
- Alert on impact, not on every metric. Ask whether a person would need to do something. If not, put it on a dashboard instead.
- Use duration windows. Require the condition to hold for 10 to 15 minutes so brief spikes do not fire.
- Use dynamic thresholds where load follows a daily pattern.
- Route by severity. Critical alerts reach a phone, warnings create tickets, informational alerts go to a daily summary.
- Give every alert an owner and a first troubleshooting step in its description.
- Suppress planned noise with alert processing rules during maintenance windows.
- Review monthly. Delete or tune any alert that fired and led to no action.
What Happens After an Azure Monitor Alert Fires?
This is where Azure monitoring stops being a technical feature and becomes an operational job. Azure Monitor detects high utilization, fires the alert and tells the IT team. Then someone needs to:
- Review the alert and acknowledge it
- Decide whether it represents a genuine issue
- Investigate the underlying cause
- Check related resources and services
- Troubleshoot the workload
- Decide whether a configuration or capacity change is needed
- Implement the fix
- Keep watching the environment afterward
For a company with a dedicated Azure operations team, that is a daily routine. For a small business with a handful of IT staff juggling many other tasks, giving each alert that attention consistently, at 2 a.m. and on weekends, is hard.
This is the gap managed Azure support fills: not a replacement for Azure Monitor, but the people and process that turn its alerts into fixes.
How Much Does Azure Monitor Cost?
Azure Monitor is billed by what you use, and the bill is dominated by one thing: log data. These are the parts that cost nothing:
| Free | What you get |
|---|---|
| Platform metrics | Collected and available at no charge |
| Activity log | Available for 90 days at no charge |
| Log ingestion | First 5 GB per month per billing account on pay-as-you-go |
| Log retention | First 31 days included |
| Metric alerts | First 10 monitored metric time series per month |
| Notifications | First 1,000 emails per month, and the first 100 SMS messages for the United States |
Beyond that, these are the main charges. Prices below are Microsoft’s pay-as-you-go list prices for the East US region as published in November 2026, so confirm current figures for your region before budgeting:
| Item | List price |
|---|---|
| Analytics Logs ingestion | $2.30 per GB |
| Basic Logs ingestion | $0.50 per GB |
| Auxiliary Logs ingestion | $0.05 per GB |
| Metric alert, each time series beyond the free 10 | $0.10 per month |
| Log search alert rule | $0.50 (15-minute), $1.00 (10-minute), $1.50 (5-minute) or $3.00 (1-minute) per month |
| SMS notification (United States) after the free 100 | About $0.006 each |
A quick illustration makes the pattern clear. If a workspace ingests 30 GB in a month, the first 5 GB are free and the remaining 25 GB cost $57.50. Ten log search alert rules evaluated every 5 minutes add $15. The alerts are the cheap part. The logs are where costs creep, because a busy server can send far more data than anyone expects.
To keep log costs in check:
- Collect only the tables and counters you will actually use, and filter at collection with the data collection rule.
- Send high-volume, low-value logs to Basic or Auxiliary tiers when you do not need to alert on them.
- Check the Usage table in Log Analytics monthly to see which data types ingest the most.
- Consider a commitment tier once ingestion is steady at 100 GB a day or more, and use a daily cap with care, because it stops collection when reached.
Monitoring cost belongs in the wider cost conversation. Our guide to right-sizing and cost controls in Azure covers the rest of the bill, and automating routine cost checks shows how to keep it from drifting.
Common Azure Monitoring Mistakes
- Assuming monitoring is on. Platform metrics are collected, but memory, disk and logs are not until you set them up.
- Sending alerts to one person. When they are away or leave, the alerts go nowhere.
- Never testing the notification path. An untested action group is a guess.
- Collecting every log forever. It inflates the bill and buries useful data.
- Ignoring Service Health and backups. Your team troubleshoots a problem Azure had already announced, or finds a failed backup only when it needs the restore.
Do It Yourself or Get Managed Monitoring?
Plenty of teams run Azure Monitor themselves, and it works well when the people and the time are there. The question is whether your team can cover the whole chain from setup to response, including out of hours.
Doing it yourself works when you have a trained on-call rotation, someone who owns the alerts and reviews them monthly, and a short outage is tolerable. Managed monitoring helps when your IT team is small and works business hours, monitoring is nobody’s main job, or downtime costs real revenue.
If the second description sounds familiar, the choice is usually between hiring and outsourcing. An independent review of your Azure setup can show what you need before you commit to either.
How Apps4Rent Helps With Azure Monitoring
Azure Monitor gives you the platform. The real value comes from connecting monitoring to action: knowing what to watch, which alerts matter and who responds.
That is the job of our Managed Azure services. Every Managed Azure plan includes 24/7 monitoring, 24/7 support and VM and OS-level support, with real-time alerting so incidents are caught and worked before they turn into downtime your users notice. The Basic plan starts at $29.95 per server per month. The Advanced plan, at $49.95 per server per month, adds critical OS service monitoring, proactive patching and hardening, and a priority response queue. Azure consumption is billed at Microsoft’s own pricing, as a separate line from the management fee.
We work alongside Azure Monitor, not instead of it. Beyond monitoring, our engineers handle VM management, OS patching, security hardening, backup, disaster recovery and cost optimization, and you reach them by phone, chat or email. If your team has Azure workloads but not the time to watch them around the clock, contact us.
Not Sure Which Alerts Matter for Your Environment?
Most alert problems come from the same two mistakes: too many rules nobody reads, and the one rule that mattered was never created.
Tell us what you run in Azure. An Apps4Rent specialist will review your monitoring setup, show you the gaps and suggest what to alert on first.
Frequently Asked Questions
-
What is Azure Monitor?
Azure Monitor is Microsoft’s built-in monitoring service for Azure, hybrid and multicloud environments. It collects metrics, logs and events from your resources, lets you analyze them with tools such as Metrics Explorer and Log Analytics, and sends alerts or triggers automated actions when conditions you define are met.
-
Is Azure Monitor free?
Partly. Platform metrics and the activity log are collected at no charge, and the first 5 GB of log ingestion per month is free. Beyond that you pay for log ingestion, alert rules and some notifications. Log data is usually the largest cost.
-
Does Azure Monitor prevent downtime?
No. Azure Monitor cannot stop a failure from happening. It gives you visibility and can notify your team, or start an automated response, when a condition suggests a problem. A person or an automation still has to investigate and act on the alert.
-
Does Azure monitoring only apply to virtual machines?
No. You can monitor virtual machines, applications, databases, storage accounts, networking resources, containers and many other Azure services.
-
What is the difference between Azure Monitor and Log Analytics?
Azure Monitor is the overall service. Log Analytics is the part of it that stores log data in a workspace and lets you query that data with Kusto Query Language (KQL). Azure Monitor also includes metrics, alerts, workbooks and insights that sit alongside Log Analytics.
-
Why can’t I see memory metrics for my Azure VM?
CPU, disk operations and network figures are platform metrics collected automatically from the host. Memory and logical disk space are measured inside the guest operating system, so they need the Azure Monitor Agent and a data collection rule. VM insights can install both for you.
-
What is the difference between Service Health and Resource Health?
Service Health reports problems on Azure’s side that affect the services and regions you use, such as outages, planned maintenance and advisories. Resource Health reports the health of one specific resource, such as a single virtual machine, and shows whether the cause is Azure or something in your configuration.
-
What happens if you have too many Azure alerts?
Alert fatigue sets in. Teams start ignoring notifications and the one that matters gets lost. Review thresholds, remove duplicates, route alerts by severity and delete any alert nobody acts on.
-
Is Azure monitoring enough to manage an Azure workload?
No. Monitoring tells you something needs attention. Managing a workload also takes troubleshooting, patching, security administration, right-sizing, backup and ongoing operational support.
-
Does Apps4Rent replace Azure Monitor?
No. Apps4Rent works with Azure Monitor, not instead of it. Our Managed Azure plans add 24/7 monitoring and support, real-time alerting and engineers who respond to what the alerts find, along with patching, security, backup and cost control for your environment.
Looking for help with Azure?
Our Azure experts can help you.
24/7 Azure Monitoring and Support
Stop Finding Out About Problems From Your Customers
Tell us what runs in your Azure environment. We will show you what should be monitored, which alerts are worth having and how our engineers respond when one fires.
Or call 1-866-716-2040 to talk it through first
Need 24/7 Azure monitoring without hiring for it?