14.8.3: Hands On Lab Exercises

Lab: Designing a Future-Ready Cloud Architecture

Objective: Design a cloud architecture that incorporates emerging trends: serverless, AI/ML, edge computing, and sustainability practices.

Scenario: SmartCity Transit is building a next-generation public transportation platform for a city of 2 million people. Requirements:

• Real-time tracking of 500 buses and 200 trains via IoT GPS sensors
• AI-powered demand prediction to optimize routes and schedules
• Passenger mobile app with real-time arrival predictions (response time < 200ms)
• Computer vision for crowding detection at 100 transit stations (cameras)
• Natural language chatbot for passenger inquiries (powered by a foundation model)
• Historical analytics for city planners (ridership trends, peak times, route efficiency)
• Target: carbon-neutral cloud operations
• Budget: $50,000/month cloud infrastructure

Instructions:

1. Architecture Design (25 min): Design a comprehensive architecture incorporating:
   • Edge computing for real-time vehicle tracking and station cameras
   • Serverless components for event-driven processing
   • AI/ML pipeline for demand prediction and route optimization
   • Foundation model integration for the chatbot
   • Data pipeline for historical analytics Create a diagram showing all components and data flows.
2. AI/ML Pipeline Design (20 min): Design the ML pipeline for demand prediction:
   • Data sources (ridership data, weather, events, time of day)
   • Feature engineering and feature store
   • Model training approach (how often, what compute resources)
   • Model serving for real-time predictions (serverless vs. dedicated GPU)
   • Monitoring for model drift and retraining triggers
3. Serverless vs. Container Decisions (15 min): For each major component, decide whether to use serverless (Lambda/Cloud Functions), serverless containers (Fargate/Cloud Run), or traditional containers (EKS/GKE). Create a decision matrix: Component | Compute Model | Justification | Cold Start Acceptable?
4. Sustainability Plan (15 min): Design the sustainability approach:
   • Region selection based on renewable energy availability
   • Carbon-aware scheduling for batch workloads (model training, analytics)
   • Right-sizing and auto-scaling to minimize idle resources
   • Monitoring and reporting carbon footprint
   • Estimate vs. target: how do you measure whether you're carbon-neutral?
5. Cost Optimization (15 min): Create a monthly cost estimate within the $50K budget. Identify the top 3 cost drivers and propose optimization strategies. How does the choice of serverless vs. containers vs. GPU instances affect the total cost?

Deliverables:

• Comprehensive architecture diagram
• AI/ML pipeline design document
• Serverless vs. container decision matrix
• Sustainability plan with specific practices
• Cost estimate with optimization strategies