GCP PCA Case Study Complete Guide: Mountkirk Games / EHR Healthcare / HRL / TerramEarth Deep Dive

The 4 case studies are the defining feature of the PCA exam. They test your ability to read each company's requirements, constraints, and future plans, then design the optimal GCP architecture. This article organizes each case by industry, challenge, and recommended architecture, and summarizes the frequent patterns and the reasoning behind correct answers.

Case 1: Mountkirk Games

Recommended Architecture

• Game backend: GKE Autopilot (multi-region), Global HTTPS LB
• DB: Spanner (global consistency) + Memorystore Redis (real-time leaderboards)
• Streaming: Pub/Sub → Dataflow → BigQuery (player behavior analytics)
• ML: Vertex AI for churn prediction and dynamic difficulty tuning
• CI/CD: Cloud Build + Cloud Deploy + Binary Authorization

Case 2: EHR Healthcare

Recommended Architecture

• Identity federation: Cloud Identity + Active Directory Federation (SAML/SCIM)
• Security: CMEK, Cloud HSM, VPC Service Controls, Access Transparency
• Hybrid: Dedicated Interconnect, Anthos / GKE on VMware
• Observability: Cloud Monitoring + Cloud Audit Logs → BigQuery export
• BAA: Use only HIPAA BAA-covered services (Compute / Storage / BigQuery / Healthcare API)

Case 3: Helicopter Racing League (HRL)

Recommended Architecture

• Delivery: Media CDN + Cloud Storage + Transcoder API
• Prediction ML: Vertex AI (real-time winner prediction), BigQuery ML
• Streaming: Pub/Sub → Dataflow → BigQuery → Looker
• UX: Global External HTTPS LB, Cloud CDN caching
• DR: Multi-region Cloud Storage, Spanner

Case 4: TerramEarth

Recommended Architecture

• IoT ingestion: Pub/Sub (global endpoint)
• Processing: Dataflow (Apache Beam) streaming
• Time-series storage: Bigtable (low-latency reads/writes)
• Analytics: BigQuery + Looker (dealer-facing dashboards)
• Predictive maintenance ML: Vertex AI anomaly detection, AutoML Tables
• Legacy migration: Migrate to Containers / Migrate to Virtual Machines

Common Recurring Patterns

• Need global consistency → Spanner; Cloud SQL HA + Read Replicas are region-scoped only
• High-volume time-series / IoT data → Bigtable + Pub/Sub + Dataflow
• Streaming analytics → Pub/Sub → Dataflow → BigQuery is the standard pattern
• HIPAA / regulatory compliance → CMEK + VPC Service Controls + Audit Logs export
• Multi-cloud / hybrid → Anthos / GKE Enterprise + Dedicated Interconnect
• CI/CD with security → Cloud Build + Binary Authorization + Artifact Registry

Study Approach

1. Read each case study in 30 minutes and turn the requirements, constraints, and future plans into a mind map
2. List 10 GCP products likely to come up in each case
3. Solve 50 case study questions on practice platforms (Udemy / Whizlabs)
4. Analyze every wrong answer and tie it back to the relevant part of the case
5. One week before the exam, re-read each case and redesign your recommended architecture on a single page

Google Cloud is a trademark of Google LLC. This article compiles independent study information and is not affiliated with Google LLC. The case study contents may be revised officially, so please check the latest version on the official Google Cloud site before your exam.