Prepare and test your skills
Prepare and test your skills
Free sample question, answered and explained below. The Professional Cloud DevOps Engineer practice bank has 1,455 more like it. Try 10 free: instant score, no account.
A financial services organization runs a high-throughput microservices application on Google Kubernetes Engine (GKE) processing over 50,000 requests per second. The DevOps engineering team is experiencing excessive Google Cloud Trace ingestion costs while attempting to monitor Service Level Objectives (SLOs).
The team has established the following operational requirements:
Which OpenTelemetry tracing and sampling strategy should the team implement?
This strategy uses the OpenTelemetry spanmetrics processor to aggregate span data into request count, duration, and error metrics, discarding all underlying span data before exporting to Cloud Trace.
The organization only required high-level aggregate SLO dashboards without the ability to inspect individual transaction waterfalls.
The OpenTelemetry Collector tail_sampling processor evaluates sampling decisions at the end of a trace lifecycle rather than at the beginning. By deploying the OpenTelemetry Collector in a gateway (cluster-level) architecture, spans from across all downstream microservices are routed and assembled into complete trace graphs before determining whether to export them to Cloud Trace or discard them.
status_code filter (e.g., ERROR, STATUS_CODE_ERROR) to retain 100% of spans associated with failed requests.latency filter policy inspects the total trace duration and automatically retains traces exceeding the 1,500 ms threshold for end-to-end latency analysis.probabilistic policy configured at 1% (0.01), effectively dropping 99% of routine telemetry before transmission to Google Cloud Trace.Head-based sampling occurs at span initialization before the outcome or duration is known. Tail sampling in the OpenTelemetry Collector gateway is the industry standard for enforcing fine-grained, outcome-based filtering rules that balance diagnostic depth against Cloud Trace ingestion quotas.
This approach attempts to use client-side SDK head-based sampling (TraceIdRatioBased) configured directly inside each application workload, attempting to retroactively adjust span sampling flags when an exception occurs.
isSampled = false), downstream services following standard W3C traceparent propagation will inherit the unsampled decision. Retroactively changing span attributes upon an uncaught exception in a child service will result in orphaned, incomplete trace fragments rather than full distributed traces.The goal was simply uniform random sampling where transaction duration and final outcome were irrelevant to the sampling decision.
This option ingests all spans into Google Cloud Trace and attempts to use Cloud Logging exclusion filters to drop 99% of normal traces after receipt.
WriteSpans API. Sending 100% of trace data directly into Cloud Trace incurs full ingestion fees immediately upon ingestion.resource.type, logName), not distributed trace spans.The telemetry in question consisted of application text logs routed to Cloud Logging rather than distributed tracing spans routed to Cloud Trace.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated Validated