Ensuring solution and operations reliability
Free Google Cloud Certified - Professional Cloud Architect practice — 6 questions on Ensuring solution and operations reliability, with explanations. No sign-up. Full 12-question mixed test →
A company has an SLO of 99.95% availability for its public API measured over a rolling 28-day window. Monitoring shows the current burn rate will exhaust the entire error budget in 3 days. The product team wants to ship a major new feature this week that adds significant new code paths to the same service. What should the architect recommend?
This is the standard SRE error-budget policy: when the burn rate indicates the budget will be exhausted, feature launches that add risk are paused and engineering effort shifts to reliability work until the budget recovers. This protects users while still allowing normal release velocity once risk decreases.
A critical OLTP workload currently runs on a single-region Cloud SQL for PostgreSQL instance. The business now requires an RPO of 5 minutes and an RTO of 15 minutes in the event of a full regional outage, with minimal ongoing operational overhead. Which approach best meets these requirements?
Cloud Spanner multi-region configurations use synchronous, Paxos-based replication across regions, providing near-zero RPO and automatic failover with minimal manual intervention, comfortably meeting a 5-minute RPO and 15-minute RTO with low ongoing operational burden.
A microservices application running on GKE experiences intermittent high latency on certain user requests. The team needs to visualize the full request path across multiple services and pinpoint exactly which downstream service call is adding the delay. Which tool should they use?
Cloud Trace is purpose-built for distributed tracing: it captures per-request latency data as a request flows through multiple services and renders the call graph so the specific slow hop can be identified directly.
A team runs several services on App Engine and Cloud Run. They want unhandled exceptions from all services automatically aggregated and grouped by root cause (deduplicating identical stack traces) and to be notified when a new error type first appears. Which Google Cloud tool is designed specifically for this?
Error Reporting automatically aggregates and groups exceptions across App Engine, Cloud Run, GKE, and Compute Engine by analyzing stack traces, deduplicates recurring errors, and can notify when a new error group is first detected — exactly matching the requirement.
An architect needs to configure a canary release for a service running on GKE, routing approximately 5% of traffic to the new version. No service mesh (such as Istio/Anthos Service Mesh) is installed, and native Kubernetes objects must be used. Which configuration achieves this?
Without a service mesh, the standard native-Kubernetes canary pattern is to create two Deployments with identical pod labels matched by a single Service's selector. Since a Service load-balances round-robin across all matching pod endpoints, controlling the ratio of stable-to-canary replica counts (e.g., 19 stable pods and 1 canary pod) approximates the desired traffic percentage.
A production outage was caused by an engineer making a manual configuration change directly in production without following the documented change-management process. During the post-incident review, which action best aligns with Google's SRE blameless postmortem culture?
Blameless postmortems focus on identifying and fixing systemic weaknesses (e.g., lack of automated guardrails, insufficient change review enforcement) rather than assigning individual blame, since punishing individuals discourages honest reporting and doesn't prevent recurrence; the goal is durable, process-level fixes.
Ready for the real thing?
The full course: two full-length practice tests, video lessons for every exam domain, hands-on labs and detailed explanations.
undefined $34.99 with code A5F51EAB51242FE60905 — valid through Sep 23.
Get my $34.99 deal →