Prepare and test your skills
Prepare and test your skills
Managing incident lifecycles and service health means deciding quickly whether a failure comes from your own code or from the Google Cloud platform, then tracking the response in one place. Personalized Service Health (opens in a new tab) filters platform disruptions to the projects you own, while the public Google Cloud Service Health dashboard shows status for all customers. Together with organization views, application views, and alerts, these tools let you evaluate the blast radius and coordinate the next action.
Public and personalized service health answer two different questions about a platform disruption. Google Cloud Service Health is the public dashboard at status.cloud.google.com that reports service status by region (opens in a new tab). Personalized Service Health is the filtered view in the Google Cloud console at Service Health > Google Cloud Incidents that shows only disruptions relevant to your projects.
Personalized Service Health lets you filter incidents by service, location, relevance, and status. Each incident detail shows scope of impact, symptoms, available workarounds, and progress updates on resolution. Use the public dashboard to check whether Google Cloud has a known regional or global event, then use Personalized Service Health to confirm whether your projects are in scope and whether to wait, apply a workaround, or fail over.
Exam tip: Choose the public Service Health dashboard for all-customer status by region, and choose Personalized Service Health for incidents filtered to your projects.
Organization and application monitoring aggregate platform incidents so you do not have to check each project separately. Organization Views provides a unified view of active incidents affecting any project in your Google Cloud organization, including which services and projects are impacted. The Service Health API provides the same organization-level events programmatically, including events that might be impacting or have impacted the organization.
Application-level monitoring connects platform incidents to applications you manage in App Hub. Cloud Hub displays Google Cloud incidents for your App Hub applications when the incident relates to products that support App Hub and Gemini Cloud Assist and the incident has Related relevance. To display incidents in Cloud Hub when no application exists yet, complete the setup in order:
servicehealth.events.list plus apphub.applications.get on the application and apphub.applications.list.The three health views differ by scope and use:
| View | Scope | Choose when |
|---|---|---|
| Google Cloud Service Health | All customers, by service and region | You need to confirm a suspected regional or global platform disruption |
| Personalized Service Health | Incidents relevant to your projects | You need to evaluate impact to your workloads and coordinate response |
| Organization Views and Service Health API | All projects in an organization | You need an aggregated, cross-project or programmatic view |
Configuring alerts and notifications for service health ensures the on-call team learns about disruptions without polling the dashboard. Service Health logs service health events in Cloud Logging, and Cloud Logging delivers those events to alerting. You can create alerts for when new incidents are reported, when existing incidents are updated, or when incidents for specific Google Cloud products or locations are created or updated.
Notification routing depends on Essential Contacts, which customizes who receives Google Cloud notifications instead of relying only on default Identity and Access Management (IAM) role recipients. Essential Contacts are managed in the console at IAM & Admin > Essential Contacts for a selected project, folder, or organization. Contacts set at an organization or folder do not automatically show child-resource contacts, and child resources inherit ancestor contacts that the viewer has permission to see.
Enabling Personalized Service Health before an incident ensures the dashboard and alerts are ready before peak traffic or a planned event. Personalized Service Health depends on the Service Health API being enabled ahead of time in the selected project. Enabling APIs requires the serviceusage.services.enable permission, which is included in the Owner role (roles/owner) or the Service Usage Admin role (roles/serviceusage.serviceUsageAdmin).
Preparation also requires access and communication readiness for a peak capacity event (opens in a new tab). A centralized incident management system tracks the incident through its lifecycle, while a communication plan defines roles for incident commanders, investigators, and communicators. Teams must also confirm they can create support cases, troubleshoot issues, and access the disaster recovery environment before the incident occurs.
GKE service incident insights connect an ongoing Google Cloud incident to the specific Google Kubernetes Engine (GKE) clusters it affects. Insights for the GKE_RELIABILITY_INCIDENT subtype are available through the console, the Google Cloud CLI, or the Recommender API. Each insight names the impacted cluster and provides an incident identifier for Cloud Customer Care (opens in a new tab), a description from the incident response team, the last effective update time, and a recommended mitigation action when one exists.
A GKE service incident insight remains visible until the Google Cloud incident response team mitigates the incident and determines the insight is no longer relevant. There is a delay between the time the incident stops impacting resources and the time the insight is removed. If you applied a workaround and no longer want to see the insight, you can dismiss it.
Post-deployment maintenance and optimization is the ongoing work that keeps a deployed Google Cloud solution reliable, performant, and cost-efficient after go-live. The work centers on three habits: right-sizing resources to match actual demand, keeping platforms patched and current, and regularly validating that disaster recovery procedures still work. Cloud Monitoring, Recommender, autoscaling, and structured resilience and disaster recovery testing provide the feedback loop that connects usage data to action.
Right-sizing is the practice of continuously monitoring utilization and adjusting allocation so resources match actual demand. Over-provisioning creates unnecessary cost, while under-provisioning creates performance bottlenecks that affect application performance and user experience. The balance requires proactive review rather than a one-time sizing decision at deployment.
Cloud Monitoring provides real-time visibility into utilization metrics so teams can track usage patterns and spot inefficiencies. Data flows from running resources to Cloud Monitoring, and Recommender analyzes that utilization data to produce intelligent recommendations for optimizing allocation. Teams can add custom metrics to track workload-specific signals and configure alerts that notify administrators when predefined thresholds are met, so administrators can adjust allocation before cost or performance degrades.
Exam tip: Cloud Monitoring shows what resources are using, while Recommender suggests how to resize them — use Monitoring to detect the pattern and Recommender to decide the new size.
Autoscaling and continuous optimization adjust capacity and architecture as workload needs change. Autoscaling dynamically adjusts capacity based on workload fluctuations so applications have resources when needed without paying for idle over-provisioning. The autoscaling policy defines the minimum and maximum capacity and the conditions that trigger scaling, adding capacity when demand rises past the threshold and removing capacity when demand falls.
Compute Engine managed instance groups (MIGs) are groups of virtual machines (VMs) that are managed and scaled as a single entity. A MIG autoscaling policy specifies the minimum and maximum number of VMs to maintain and the trigger condition, for example adding VMs when CPU utilization reaches a certain threshold and removing VMs when utilization drops below a different threshold. This model depends on the workload signal: the MIG depends on Monitoring metrics, and the metrics drive the add-or-remove decision.
Optimization also means reassessing needs over time as part of continuous improvement (opens in a new tab). As business evolves, user behavior, traffic profiles, and dependencies change, so teams must periodically review utilization and costs and determine if architectural changes are required. Identifying critical periods is part of this review, such as seasonal shopping peaks, audit days, tax season, or trading hours, and planning adequate capacity, testing against peak-load conditions, and freezing production code during those windows supports scalable and resilient apps (opens in a new tab).
Validating disaster recovery and resilience procedures proves that recovery plans and resilience designs behave as expected before a real outage. Teams should schedule and conduct disaster recovery testing using the disaster recovery planning guide, then review the test results to determine if architectural changes or resource optimizations are required. If additional help is needed, the account team can discuss best practices for disaster recovery and incident response processes.
Resilience testing to ensure operational readiness and performance using CloudOps (opens in a new tab) uses a structured escalation where basic failures are introduced first and then made more severe, with behavior, alerts, and metrics validated and documented at each stage:
This ordered sequence moves from partial disruption to full loss, so each stage confirms a stronger level of independence before the next stage removes more of the environment.
Applying maintenance and staying current keeps platforms secure, supported, and aligned with current best practices. Teams use Unified Maintenance setup guidance to coordinate maintenance work and stay up to date with the latest Google Cloud advancements, features, and best practices to maximize service capability. Keeping systems current reduces the risk that outdated components cause failures or block scaling during critical periods.
Automation makes maintenance repeatable and reduces configuration drift. Infrastructure as code (IaC) stores provisioning and configuration logic in source control so changes are versioned, tested, and deployed through continuous integration and continuous deployment (CI/CD) pipelines instead of manual steps. Immutable infrastructure extends this idea by redeploying resources from the updated configuration rather than modifying running resources, which leads to more predictable deployments and rollbacks.
Diagnosing system issues using Cloud Logging (opens in a new tab) and Cloud Monitoring (opens in a new tab) is the practice of collecting logs and metrics, watching them for abnormal behavior, and using them to find the root cause of failures. Log data flows from designated infrastructure and application sources to centralized collection, where dashboards and analytical tools make performance and health visible. Metrics and log events then feed alerting policies, and alerts lead investigators back to the same centralized logs and metrics for diagnosis and repair.
Support of a deployed solution follows an ordered operational sequence:
A left-to-right process flow showing centralized log and metric collection feeding monitoring and analysis, then alerting, then troubleshooting and remediation with a return loop to healthy service.
In this sequence, monitoring depends on collection from sources, alerting depends on monitoring signals, and troubleshooting depends on both logs and metrics preserved from the earlier stages.
Monitoring performance, health, and security events in Cloud Monitoring is continuous observation of system performance, application health, and security events through dashboards, visualizations, and analytical tools. Log and metric data flows from designated sources into centralized views, so operators can compare current behavior against normal patterns in one place. This centralized visibility supports detection of performance bottlenecks, degraded application health, and security-relevant activity before users are broadly affected.
Defining alerting policies to trigger notifications turns monitoring signals into action. An alerting policy applies to a metric or log-based signal, evaluates incoming data against its condition, and sends a notification from the alerting pipeline to a configured channel when the condition is met. Because the notification carries the timing and scope of the violation, responders can move quickly from detection to investigation of the affected infrastructure or application.
Troubleshooting and remediating with logs and metrics is the use of collected logs and metrics to investigate, diagnose, and resolve production issues while minimizing downtime. Investigation flows from an alert or reported symptom to centralized logs and detailed performance metrics, where correlated entries reveal the failing component and root cause. Resolution then flows from diagnosis back to the affected system, restoring full operation and reducing the risk of extended outage. This step-by-step troubleshooting (opens in a new tab) approach relies on evidence preserved from monitoring and alerting, which is part of the shared responsibility (opens in a new tab) for operating production systems and getting support (opens in a new tab) when needed.
Centralized observability with Cloud Logging and Cloud Monitoring gives operations teams broad benefits for reliability, security, and efficiency. The shared log and metric foundation connects detection, alerting, and diagnosis, so each stage strengthens the next.
Together, these benefits allow teams to find bottlenecks faster, resolve failures with evidence, and keep production systems stable and compliant.
Google Cloud Service Health is the public dashboard at status.cloud.google.com that reports service status by region for all customers, while Personalized Service Health is the filtered view in the Google Cloud console at Service Health > Google Cloud Incidents that shows only disruptions relevant to your projects. Use the public dashboard to check whether Google Cloud has a known regional or global event, then use Personalized Service Health to confirm whether your projects are in scope and whether to wait, apply a workaround, or fail over.
Organization Views provides a unified view of active incidents affecting any project in a Google Cloud organization, including which services and projects are impacted. The Service Health API provides the same organization-level events programmatically, including events that might be impacting or have impacted the organization.
Cloud Monitoring provides real-time visibility into utilization metrics so teams can track usage patterns and spot inefficiencies, while Recommender analyzes that utilization data to produce recommendations for optimizing allocation. Data flows from running resources to Cloud Monitoring to Recommender, so Monitoring detects the pattern and Recommender helps decide the new size.
An enterprise runs a suite of business-critical production workloads on Compute Engine virtual machines. The operations team needs to establish a post-deployment maintenance strategy that satisfies the following requirements:
Which configuration should you implement on the Compute Engine instances?
Last updated