Watching the sky

Azure Advisor

Azure Advisor reviews your Azure resources and gives personalized, actionable recommendations, in the portal or through an application programming interface (API), and can notify you when new ones appear. Recommendations fall into five categories:

• Reliability: keep apps running, for example by adding zone redundancy
• Security: recommendations from Defender for Cloud
• Performance: make apps faster
• Operational excellence: better processes, workflows and deployment practices
• Cost: cut waste, such as idle virtual machines (VMs) or unused disks

Exam tip Advisor = 'what should I improve?' It recommends; it doesn't track live outages or change resources by itself.

Azure Service Health

Azure Service Health tells you about the health of Azure itself, through three views that go from wide to narrow:

• Azure status: every Azure service in every region, worldwide; the public status page
• Service Health: just the services and regions you use: service issues (outages), planned maintenance, and health and security advisories. You can set alerts on these
• Resource Health: the health of your individual resources, such as one VM, including its past availability

Mr. Barkley: So Service Health answers 'was it Microsoft?'

Dev: And it keeps a history, handy for writing up incidents.

Exam tip Microsoft outage in my region = Service Health. Is my one VM healthy = Resource Health. Everything, everywhere = Azure status.

One platform for metrics and logs

Azure Monitor collects, analyzes and acts on monitoring data from Azure, on premises and other clouds. It gathers two main kinds of data: metrics (numbers over time, like central processing unit (CPU) percentage) and logs (records of events, like errors and sign-ins). Data comes from applications, operating systems, Azure resources, subscriptions and the tenant. You view it in charts, dashboards and workbooks.

Log Analytics

Log Analytics is the tool in the Azure portal where you write and run queries against log data that Azure Monitor collects into a Log Analytics workspace. Queries use Kusto Query Language (KQL). Run a quick filter, or dig deep and chart the results.

AppRequests
| where Success == false
| summarize count() by bin(TimeGenerated,5m)

Azure Monitor alerts

Azure Monitor alerts tell you when a condition is found in monitoring data, so you can act before users notice. Alerts can watch metrics (CPU above 85% for 10 minutes), log search results, or the activity log (someone deleted a VM). When an alert fires, an action group decides what happens: email, text message, push or voice notification, or automation such as Azure Functions, Logic Apps or a webhook.

Priya: So next time my phone buzzes before the customers' phones do?

Dev: That's the whole point of an alert.

Application Insights

Application Insights, a feature of Azure Monitor, is application performance monitoring (APM) for apps in Azure, on premises or in another cloud. It tracks request rates, response times and failure rates, calls to dependencies like databases and APIs, page views and load times. It can also send synthetic test requests (availability tests) to check the app responds even when nobody is using it.

Exam tip Monitor = the platform. Log Analytics = query the logs (KQL). Alerts + action groups = tell me and act. Application Insights = how is my app doing.

The exam at a glance

Microsoft Azure Fundamentals (AZ-900) is a 45-minute exam; Microsoft says most of its exams have 40 to 60 questions. Scores run from 1 to 1,000 and you pass at 700, a scaled score rather than 70% of the questions. Three skill areas: cloud concepts (25-30%), Azure architecture and services (35-40%), and Azure management and governance (30-35%). No hands-on labs.

You: So a weak area can be balanced by a strong one?

Dev: Right. You pass on the whole score, not per section.

Exam tip There's no penalty for guessing, and multi-part questions usually give a point per correct part. Never leave anything blank.

Question types and pacing

Expect multiple choice, multiple response ('Choose two'), drag-and-drop matching and series of yes/no statements. That's under a minute per question, so keep moving.

• Read the last line first: what is it really asking?
• Watch for deciding words: 'least administrative effort', 'most cost-effective', 'must survive a regional outage'
• In yes/no series, judge each statement on its own; some series can't be revisited once you move on
• Mark tough questions for review and come back with the time left
• Try Microsoft's free practice assessment and the exam sandbox before test day

Look-alike pairs

Most wrong answers on AZ-900 are the right tool for a different job. The pairs to keep apart:

• Azure Policy (resource rules) vs Azure RBAC, role-based access control (what users can do) vs resource locks (stop deletes)
• Advisor (recommendations) vs Service Health (Azure's outages) vs Monitor (your telemetry)
• Pricing calculator (planned Azure) vs TCO (total cost of ownership) calculator (on premises vs Azure) vs Cost Management (actual spend)
• Blob (objects by URL) vs Files (SMB/NFS shares); GRS (failover first) vs RA-GRS (read anytime)
• Entra ID (cloud identity) vs Entra Domain Services (managed domain); Arc (manage in place) vs Migrate (move)

Ranger: paws at the Policy vs RBAC line twice

You: Yes, Ranger. That one's on every practice test.