Operating a cloud environment means much more than just keeping servers running. "Are my servers actually working correctly right now?" "Are there areas where money is being wasted?" "Is this an Azure problem, or is it my code?" "Why was response time slow at 3 AM yesterday?" You need answers to all of these questions to truly operate the cloud well.
AZ-900 monitoring questions focus on clearly distinguishing the roles of three services: Azure Advisor, Azure Service Health, and Azure Monitor. Each one answers a different question. This post explains all three using everyday analogies and concrete examples anyone can follow.
Why Is Monitoring Needed?
Imagine a modern car. The dashboard has a speedometer, fuel gauge, engine temperature indicator, battery level, and warning lights. What if there was no dashboard at all? You would not know the fuel was running out until the engine sputtered to a stop in the middle of the highway. You would have no way to detect an overheating engine before it seized.
Cloud environments work the same way. Operating without monitoring means: CPU exceeding 90% and slowing your service — you will not know until users complain A database disk filling up — discovered only after your service crashes An Azure data center outage affecting your service — you find out late, from angry users Security threats go undetected until damage is done Unused resources burning thousands of dollars a month — invisible waste
Azure addresses all of these problems with three core monitoring services, each serving a different purpose.
Azure Advisor — Your Personal Doctor's Annual Health Report
Think of Azure Advisor as your personal physician's comprehensive annual health report. Your doctor reviews blood tests, an ECG, blood pressure readings, and cholesterol levels, then delivers proactive advice: "Your blood pressure is slightly elevated — watch your salt intake," "Your cholesterol is approaching a risky range — consider changing your diet," "It is time for your vaccination booster." The doctor advises before you get sick, not after.
Azure Advisor works exactly the same way. It continuously analyzes your Azure environment and — even if nothing is wrong right now — identifies things that could become problems and says "here is how to improve." And it is completely free.
Azure Advisor's 5 Recommendation Categories
Azure Advisor performs a comprehensive diagnosis of your Azure environment across 5 health categories.
| Category | What It Examines | Example Recommendation | Doctor Analogy | |---------|-----------------|----------------------|---------------| | Reliability | Whether services run continuously without interruption | "This VM is a single instance. Deploy across an availability set or availability zone" | Heart health — "Increase your exercise" | | Security | Security vulnerabilities and threat vectors | "This storage account allows public access. Disable it." "Accounts not using MFA detected." | Immune system — "Get your vaccination" | | Performance | Whether resources are running at optimal speed | "This VM's CPU is consistently above 90%. Scale up the size." "This SQL query has no index." | Fitness checkup — "This muscle group has weakened" | | Cost | Unnecessary spending and cost optimization opportunities | "3 VMs have been idle for 7+ days. Consider shutting them down." "Switch to reserved instances for 40% savings." | Blood pressure — "You are overeating" | | Operational Excellence | Operational efficiency and best practices | "You have not configured service health alerts." "Resources are missing required tags." | Lifestyle habits — "You are not sleeping enough" |
How to Use Azure Advisor
Search for "Advisor" in the Azure Portal to access it immediately — no configuration needed. Azure automatically runs the analysis. Click on any recommendation to see exactly which resource has the issue, what the specific problem is, and step-by-step instructions for fixing it. Some recommendations offer a one-click auto-remediation button.
Azure Advisor integrates with Microsoft Defender for Cloud (formerly Azure Security Center) to provide even richer security recommendations.
Azure Service Health — A Smart Weather App
Azure Service Health tells you about the health of Azure services themselves — not your resources, but the underlying Azure platform. Think of it as a smart weather app with three levels of granularity.
A weather app gives you three levels of information: National weather (overall conditions across the country) Your city's weather (conditions where you live) Your street's weather (conditions right where you are)
Azure Service Health provides the same three levels of detail.
Azure Service Health's 3 Components in Detail
Level 1 — Azure Status (national weather): A public page showing the overall status of every Azure service worldwide. Anyone can view it without logging in at status.azure.com. Major outages affecting Azure globally appear here.
Level 2 — Service Health (your city's weather): Shows only events affecting the specific services and regions you actually use. If you use Azure SQL Database and App Service in Korea Central, only events related to those services and that region appear. Problems in Japan East or services you do not use are filtered out — dramatically reducing noise.
Level 3 — Resource Health (your street's weather): Shows the current health status of each individual resource you have created. It answers specific questions like "Is my virtual machine named myVM-korea-01 healthy right now?" Rather than looking at service-level health, you get per-resource visibility.
Types of Events in Service Health
| Event Type | Description | Example | |-----------|-------------|---------| | Service Issues | Active, ongoing Azure service problems | "Connectivity issues with App Service in Korea Central" | | Planned Maintenance | Scheduled maintenance announced in advance | "VM host updates scheduled Feb 15, 00:00–02:00" | | Health Advisories | Service changes or feature retirement notices | "Python 3.7 runtime support ending in [Month] 2024" | | Security Advisories | Important security-related announcements | "Security vulnerability found in specific VM image" |
Setting Up Service Health Alerts
Configure alerts in Service Health to receive notifications via email, SMS, webhook, or Logic App before events affect your resources. This gives you time to prepare a response before angry users start calling — instead of scrambling to catch up after the fact.
Azure Monitor — Smartwatch Fitness Tracker
Azure Monitor is a comprehensive monitoring platform that collects, analyzes, and visualizes performance, availability, and operational data from Azure resources. Just as a smartwatch tracks heart rate, step count, sleep patterns, and oxygen saturation around the clock, Azure Monitor tracks everything from VM CPU and memory usage to web app request counts, error rates, and response times.
Two Data Types Azure Monitor Collects
Metrics: numerical measurements captured over time. "What is the current CPU percentage?" "What was memory usage over the last hour?" Metrics are lightweight, collected quickly, and ideal for visualizing as graphs and charts.
Logs: detailed event records. "At 3:12:47 AM, which user accessed which resource?" "When an error occurred, what exactly happened?" Log data is richer and more detailed than metrics, enabling deep investigation.
Log Analytics — Precision Workout Log Analyzer
Log Analytics is the dedicated tool for querying and analyzing the log data Azure Monitor collects. Just as a fitness app lets you ask "which day of last month had the highest average heart rate?" or "how many steps did I take the day after fewer than 6 hours of sleep?" — Log Analytics lets you ask precise questions about your Azure environment.
Log Analytics uses KQL (Kusto Query Language), a query language similar to SQL but purpose-built for time-series data analysis.
Real KQL query examples: Count HTTP 500 errors over the last 7 days: List VMs with CPU above 90%: Find users with many failed logins:
Azure Monitor Alerts — Emergency Smartwatch Notification
Azure Monitor Alerts automatically notify the right people when specific conditions are met. Like a smartwatch vibrating and showing a warning when your heart rate suddenly jumps above 150 bpm, Azure Monitor sends immediate alerts when configured thresholds are exceeded.
Alert types: Metric alerts: "If VM CPU utilization exceeds 90% for 5 or more minutes, send an email" Log alerts: "If there are 100+ failed login attempts in an hour, send an SMS" Activity log alerts: "If anyone deletes a resource in the production resource group, alert immediately"
Automated remediation with Action Groups: Beyond just sending notifications, alerts can trigger Azure Functions or Logic Apps to automatically fix the problem or provision additional resources — turning passive alerts into active, self-healing systems.
Alert channels: email, SMS, Azure Mobile App push notifications, webhooks, ITSM integrations (ServiceNow, etc.)
Application Insights — Dedicated App Health Monitor
Application Insights is Azure Monitor's feature for specialized monitoring of web applications and microservices. While a smartwatch tracks your entire body, Application Insights is like specialized sports performance monitoring equipment focused exclusively on your application.
Connecting Application Insights to your app is simple — add the SDK to your code, or install an agent on the server, and data collection begins automatically.
Data collected: Requests: how many HTTP requests per second are arriving? Response time: what is the average response time? What are the slowest requests? Failure rate: what percentage of requests return an error? Exceptions: which code errors are occurring, and how often? Dependencies: which external API calls and database queries are creating bottlenecks? User behavior: where