Domain 5 of 5

Monitor and maintain Azure resources

Domain · 12.5% of the AZ-104 exam

Two jobs, one rule: beyond a small default, nothing is collected or protected until you configure it

At 02:00 a VM's CPU spikes, and by morning a file it served has been deleted: the spike is a question for monitoring, the lost file a question for recovery. This domain divides into two jobs (an orienting model, not Azure terminology): Monitoring notices and explains what happens to your resources and tells the right people, and Recovery gets data back from an earlier point or keeps a workload running in another region. Both jobs follow one rule. Platform metrics are collected automatically from Azure resources and require no configuration, but guest data from inside a VM, such as performance counters and logs, isn't collected by default until a data collection rule has Azure Monitor Agent collect it. Azure Monitor collects activity log entries by default (the record of management operations such as creating a VM), but resource logs (records of operations inside a resource, such as reading a secret from a key vault) aren't collected by default and require a diagnostic setting, a setting that routes logs and metrics to a destination. An Azure VM needs backup configured and an initial backup completed before it has a recovery point to restore from. The classic trap is searching data that was never collected: the activity log doesn't typically capture read operations, so evidence of a secret read must come from resource logs, which hold it only if a diagnostic setting was collecting them at the time.

The domain unfolds in two topics, one per job

Monitor resources in Azure covers the Monitoring job: reading metrics; collecting logs with diagnostic settings and with data collection rules (DCRs), which tell Azure Monitor Agent what to collect inside a VM; querying a Log Analytics workspace with Kusto Query Language (KQL); building Azure Monitor alerts from alert rules that detect a condition, action groups that define who is notified and what actions run, and alert processing rules that change alerts after they fire; reading the Azure Monitor Insights views for VMs, storage accounts and networks; and diagnosing network paths with Azure Network Watcher. Implement backup and recovery covers the Recovery job, which splits into two services. With vaulted backup (the vault tier, which copies data to a vault, unlike snapshot-tier points, which in VM backup are stored along with the disk), Azure Backup keeps recovery points (restorable copies) in a Recovery Services vault or a Backup vault on the schedule of a backup policy. These are two distinct vault types that target different data sources, so the data source decides which you create. Azure Site Recovery replicates VMs to a different target region and fails them over there. The second topic builds on the first, because Backup Reports read diagnostics data that each vault sends to a Log Analytics workspace through a diagnostic setting you configure on the vault, and built-in Azure Backup alerts, which aren't created by an alert rule, reach people through an alert processing rule and an action group. Network security rules themselves are taught in Configure secure access to virtual networks, and storage-level protection such as blob soft delete in Configure Azure Files and Azure Blob Storage.

When two answers both work, match the tool's time frame to the requirement

Many choices in this domain pair a tool that answers about right now with one that answers over time, so first ask which time frame the requirement names. Connection troubleshoot tests a connection at a point in time instead of monitoring it over time, as Connection monitor does; the Resiliency dashboard shows backup jobs in real time, while data for the current partial day isn't shown in Backup Reports; and Azure retains activity log events for 90 days, so a longer history needs a diagnostic setting that exports them. Recovery splits on running elsewhere versus going back: Site Recovery's default policy keeps recovery points for one day and lets you choose a retained point at failover, which only sets the state the failed-over VM starts from, so getting data back from an earlier point, especially past that window, stays with Azure Backup, which keeps recovery points under a backup policy.

The two jobs of this domain and the topic that covers each

JobQuestion it answersWhat you configureDrill into
MonitoringWhat is happening to a resource, why, and who needs to know?Diagnostic settings, data collection rules, KQL queries, alert rules, action groups, alert processing rules, Azure Monitor Insights, Network Watcher toolsMonitor resources in Azure
RecoveryCan you get data back from an earlier point, or run the workload in another region?Recovery Services vaults and Backup vaults, backup policies, restores, Azure Site Recovery replication and failover, Backup Reports and backup alertsImplement backup and recovery

Decision tree

MonitoringTell someone whena condition occurs?YesAzure MonitoralertsNoExplain whathappened, or why?YesMetrics, logs,Network WatcherRecoveryNoRun it in anotherregion in an outage?YesAzure SiteRecoveryNo: restore an earlier pointAzure Backup

Subtopics in this domain