GCP Solution Architect Interview Questions¶

Comprehensive collection of interview questions with detailed answers.

Table of Contents¶

Fundamentals
Compute
Networking
Storage & Databases
Security
Architecture & Design
Scenario-Based Questions

Fundamentals¶

Q1: What are the main differences between GCP, AWS, and Azure?¶

Answer:

GCP Strengths: - Best-in-class Kubernetes (GKE) - Superior data analytics (BigQuery) - Global private network - Per-second billing - Automatic sustained use discounts - Strong in ML/AI (Vertex AI)

Key Differences: - Network: GCP uses private fiber network vs AWS/Azure public internet - Billing: GCP per-second vs AWS/Azure per-minute/hour - Discounts: GCP automatic vs AWS/Azure reserved instances - Kubernetes: GKE is native vs EKS/AKS

Q2: Explain GCP's resource hierarchy.¶

Answer:

Organization
    ↓
Folders (optional, can be nested)
    ↓
Projects
    ↓
Resources (VMs, databases, etc.)

Key Points: - IAM policies inherit downward - Billing is at project level - Projects are isolation boundaries - Organization is the root node - Folders help organize by department/environment

Example:

Acme Corp (Organization)
├── Production (Folder)
│   ├── Web App (Project)
│   └── API (Project)
└── Development (Folder)
    └── Test (Project)

Q3: What is the difference between regions and zones?¶

Answer:

Region: - Geographic location (e.g., us-central1) - Contains multiple zones - Independent failure domains - ~40+ regions globally

Zone: - Deployment area within region (e.g., us-central1-a) - Single failure domain - Low latency between zones in same region - 3+ zones per region

Best Practices: - Multi-zone for HA (99.99%) - Multi-region for global apps (99.999%) - Regional resources for HA - Zonal resources for cost optimization

Compute¶

Q4: When would you choose Cloud Run over GKE?¶

Answer:

Choose Cloud Run when: - Stateless HTTP services - Simple deployment needs - Auto-scaling to zero required - Pay-per-use pricing preferred - No Kubernetes expertise - Quick time to market

Choose GKE when: - Complex microservices - Need full Kubernetes features - Stateful applications - Custom networking required - Multi-cloud portability - Advanced orchestration needs

Example Scenario: - API Gateway → Cloud Run - Microservices Platform → GKE

Q5: Explain the difference between preemptible and spot VMs.¶

Answer:

Preemptible VMs (Legacy): - Up to 80% discount - Max 24-hour runtime - Can be preempted anytime - 30-second shutdown notice

Spot VMs (Current): - Up to 91% discount - No maximum runtime - Can be preempted anytime - Dynamic pricing - Better for batch workloads

Use Cases: - Batch processing - Fault-tolerant workloads - Dataproc secondary workers - CI/CD build agents - Data processing pipelines

Not Suitable For: - Databases - Production web servers - Stateful applications

Networking¶

Q6: Explain VPC peering vs Shared VPC.¶

Answer:

VPC Peering: - Connects two VPCs - Private RFC 1918 connectivity - No overlapping IP ranges - Transitive peering not supported - Cross-project, cross-organization

Shared VPC: - Centralized network management - Multiple projects share one VPC - Host project owns network - Service projects use network - Better for organizations

When to Use:

VPC Peering:
  • Connect different organizations
  • Temporary connections
  • Simple point-to-point

Shared VPC:
  • Enterprise environments
  • Centralized network team
  • Multiple projects, one network

Q7: How does Cloud Load Balancing work?¶

Answer:

Types:

Global HTTP(S) Load Balancer
Layer 7
Anycast IP
SSL termination
URL-based routing
Global backends
Regional Network Load Balancer
Layer 4
TCP/UDP
Regional only
Pass-through
Internal Load Balancer
Private IPs only
Regional
Layer 4 or Layer 7

Key Features: - Auto-scaling - Health checks - Session affinity - CDN integration - SSL certificates

Storage & Databases¶

Q8: When would you use Cloud Spanner vs Cloud SQL?¶

Answer:

Cloud Spanner: - Global distribution needed - Strong consistency required - Horizontal scaling - 99.999% availability - Multi-region transactions - Higher cost

Cloud SQL: - Regional application - Traditional SQL workload - Vertical scaling - 99.95% availability - Lower cost - Familiar MySQL/PostgreSQL

Decision Matrix:

Global app + Strong consistency → Cloud Spanner
Regional app + Cost-sensitive → Cloud SQL
Financial transactions → Cloud Spanner
Web application → Cloud SQL

Q9: Explain BigQuery partitioning and clustering.¶

Answer:

Partitioning: - Divides table into segments - Based on column (date, timestamp, integer) - Reduces data scanned - Improves query performance - Lowers costs

CREATE TABLE dataset.partitioned_table
PARTITION BY DATE(timestamp)
AS SELECT * FROM source;

Clustering: - Sorts data within partitions - Based on multiple columns - Co-locates related data - Further improves performance

CREATE TABLE dataset.clustered_table
PARTITION BY DATE(timestamp)
CLUSTER BY user_id, country
AS SELECT * FROM source;

Best Practice: - Partition by date/timestamp - Cluster by frequently filtered columns - Use both for optimal performance

Security¶

Q10: Explain the principle of least privilege in GCP.¶

Answer:

Concept: - Grant minimum permissions needed - Use predefined roles when possible - Create custom roles for specific needs - Regular access reviews

Implementation:

# Bad: Too broad
gcloud projects add-iam-policy-binding PROJECT \
  --member=user:dev@example.com \
  --role=roles/editor

# Good: Specific permissions
gcloud projects add-iam-policy-binding PROJECT \
  --member=user:dev@example.com \
  --role=roles/compute.instanceAdmin.v1

Best Practices: - Use service accounts for applications - Avoid primitive roles (Owner, Editor, Viewer) - Use IAM conditions for context-aware access - Enable audit logging - Regular permission reviews

Q11: How do you secure secrets in GCP?¶

Answer:

Secret Manager:

from google.cloud import secretmanager

client = secretmanager.SecretManagerServiceClient()
name = f"projects/{project}/secrets/{secret}/versions/latest"
response = client.access_secret_version(request={"name": name})
secret_value = response.payload.data.decode("UTF-8")

Best Practices: - Never commit secrets to code - Use Secret Manager for all secrets - Rotate secrets regularly - Use IAM for access control - Enable audit logging - Use Workload Identity (no keys)

Integration: - Cloud Run: Mount as env var or volume - GKE: Use Workload Identity - Cloud Functions: Access via SDK - Compute Engine: Use metadata server

Architecture & Design¶

Q12: Design a highly available web application on GCP.¶

Answer:

Architecture:

Internet
    ↓
Cloud Armor (DDoS protection)
    ↓
Global Load Balancer (HTTPS)
    ↓
Cloud CDN (Static content)
    ↓
┌─────────────────────────────┐
│  Regional GKE Cluster       │
│  (us-central1)              │
│  ┌──────────┐  ┌──────────┐│
│  │ Zone A   │  │ Zone B   ││
│  │ • Pods   │  │ • Pods   ││
│  └──────────┘  └──────────┘│
└─────────────────────────────┘
    ↓
┌─────────────────────────────┐
│  Cloud SQL (Regional)       │
│  • Primary (Zone A)         │
│  • Replica (Zone B)         │
└─────────────────────────────┘
    ↓
Cloud Storage (Multi-region)

Key Components: - Multi-zone GKE for HA - Regional Cloud SQL with replica - Global load balancer - Cloud CDN for performance - Cloud Armor for security - Monitoring and logging

SLA: 99.99% availability

Q13: How would you migrate a monolithic application to microservices on GCP?¶

Answer:

Strategy: Strangler Fig Pattern

Phase 1: Assessment - Identify service boundaries - Map dependencies - Define APIs - Plan data migration

Phase 2: Incremental Migration

Monolith → API Gateway → New Microservices
   ↓                          ↓
Old DB  ←─── Sync ────→  New DBs

Phase 3: Implementation 1. Deploy API Gateway (Cloud Endpoints) 2. Extract one service at a time 3. Deploy to Cloud Run/GKE 4. Route traffic gradually 5. Migrate data 6. Decommission old code

Tools: - Cloud Run for stateless services - GKE for complex services - Cloud SQL/Spanner for databases - Pub/Sub for async communication - Cloud Build for CI/CD

Scenario-Based Questions¶

Q14: Your application is experiencing high latency. How do you troubleshoot?¶

Answer:

Step 1: Identify the Problem

# Check Cloud Monitoring
gcloud monitoring time-series list \
  --filter='metric.type="run.googleapis.com/request_latencies"'

# Check Cloud Trace
# View distributed traces in console

Step 2: Common Causes - Database queries (use Cloud SQL Insights) - External API calls (check Cloud Trace) - Cold starts (increase min instances) - Network latency (use Cloud CDN) - Resource constraints (check CPU/memory)

Step 3: Solutions

# Add caching
gcloud redis instances create cache \
  --size=1 \
  --region=us-central1

# Optimize database
CREATE INDEX idx_user_id ON users(user_id);

# Increase resources
gcloud run services update my-service \
  --memory=2Gi \
  --cpu=2 \
  --min-instances=1

Q15: Design a real-time analytics pipeline.¶

Answer:

Architecture:

IoT Devices / Apps
    ↓
Pub/Sub (Ingestion)
    ↓
Dataflow (Stream Processing)
    ↓
┌─────────┬─────────┐
│         │         │
v         v         v
BigQuery  Bigtable  Pub/Sub
(Analytics) (Hot)   (Alerts)

Implementation:

Ingestion:

from google.cloud import pubsub_v1

publisher = pubsub_v1.PublisherClient()
topic_path = publisher.topic_path('project', 'events')

# Publish event
data = json.dumps(event).encode('utf-8')
publisher.publish(topic_path, data)

Processing:

import apache_beam as beam

(pipeline
 | 'Read' >> beam.io.ReadFromPubSub(subscription=sub)
 | 'Parse' >> beam.Map(parse_json)
 | 'Window' >> beam.WindowInto(window.FixedWindows(60))
 | 'Aggregate' >> beam.CombinePerKey(sum)
 | 'Write' >> beam.io.WriteToBigQuery(table))

Analysis:

SELECT
  TIMESTAMP_TRUNC(timestamp, HOUR) as hour,
  COUNT(*) as event_count,
  AVG(value) as avg_value
FROM dataset.events
WHERE DATE(timestamp) = CURRENT_DATE()
GROUP BY hour
ORDER BY hour DESC;

Tips for Interview Success¶

✅ Understand trade-offs between services
✅ Know pricing models
✅ Practice designing architectures
✅ Understand security best practices
✅ Know when to use each service
✅ Be familiar with gcloud commands
✅ Understand networking concepts
✅ Practice scenario-based questions

Last Updated: March 2026
Status: ✅ Complete