Google Professional Cloud Database Engineer (PCDBE) 덤프 및 해설

Your media-streaming platform uses Memorystore for Redis (Standard Tier, Redis 6.x) as a cache for user session tokens and frequently requested metadata; during live event traffic spikes, p95 cache latency jumps from 6 ms to 180–250 ms, and Cloud Monitoring shows memory utilization at 95–98% with Redis INFO reporting ~25,000 evicted_keys/min and an eviction policy of allkeys-lru; average key size is ~1.5 KB, average TTL is 5 minutes, CPU stays under 40%, and network RTT between your GKE cluster and the Redis instance is ~1 ms in the same region; you need to reduce the frequency and impact of these latency spikes. What should you do?

Your healthcare analytics company is migrating a self-managed PostgreSQL 12 database from an on-premises datacenter to Cloud SQL for PostgreSQL; after cutover, the system must tolerate a single-zone outage in the target region with no more than 3 minutes of disruption (RTO ≤ 3 minutes) and zero transaction loss (RPO = 0), and you want to follow Google-recommended practices for the migration—what should you do?

Your edtech platform uses Cloud Firestore for storage and serves a React web app to a global audience. Each day at 00:00 UTC, you publish the same Top 20 practice tips (20 documents, ~5 KB each; total payload ~100 KB) to approximately 5 million daily active users across North America, Europe, and APAC. You need to cut Firestore read costs and achieve sub-150 ms p95 load times for this daily list while the content remains identical for 24 hours. What should you do?

Your travel-tech company operates a globally distributed seat allocation platform on Cloud Spanner with 3 read-write and 6 read-only regions; after importing 12 million seat and route records from a partner, you observe write latency spikes and CPU hotspots on 2 of 8 leader replicas, and Cloud Monitoring shows hot ranges on a table keyed by a monotonically increasing ticket_id and a composite key (route_id, class) where class has only 3 distinct values; at peak 35,000 writes/second, 70% of writes concentrate in a narrow key range; to follow Google-recommended schema design practices and avoid hotspots without sacrificing strong consistency or availability, what should you do? (Choose two.)

You are building a Pub/Sub–triggered service on Cloud Functions (2nd gen) in us-central1 that must connect to a Cloud SQL for PostgreSQL instance. Your security policy requires the database to accept connections only from workloads inside the prod-vpc VPC, with no public internet exposure. The function can scale up to 200 concurrent requests during peak load, and you need stable connection management. What should you do to meet the security and performance requirements?