HipLocal Case Study

Non-italicised points represent notes taken from Google’s official case study documentation, with keywords highlighted in bold. Points in italics represent additional insights and possible solutions.

Company Overview

• Community application
  • Communication between people in close proximity
    • Streaming data: Cloud IoT Core, Pub/Sub
    • Data synchronization: Cloud Firestore
• Event planning, organizing sporting events, businesses to connect with local communities
• Initial launch in Dallas—growing into global phenomenon
  • Global and multi-regional resources

Executive Statement

• Go global—communities to span global regions
• Desire rapid growth
  • PaaS, managed services

Solution Concept

• Expand existing service
  • Update functionality
  • New locations
• Hire/train new teams to support each location in local timezone
• Ensure:
  • Smooth autoscaling
    • MIGs, Serverless
  • Provide clear uptime data
    • Cloud Monitoring
  • Analyse/respond to issues
    • Cloud Monitoring

Existing Technical Environment

• Mixture of on-prem and Google Cloud infrastructure
  • Cloud VPN, Cloud Interconnect
• Team understands app, but limited experience of globally scaled apps
  • Prefer managed solutions
• APIs run on Compute Engine instances
  • Migrate to Cloud Endpoints, Google Cloud API Gateway or Apigee
• State stored in single MySQL instance in Google Cloud
  • Migrate to Cloud SQL or Cloud Spanner
• Release cycles—dev freezes during QA testing
  • Require separate dev, QA and prod environments
  • CI/CD process with automated testing
• No consistent logging
  • Cloud Logging
  • Install logging agent on GCE instances
• Deployments manual by infrastructure engineers during quiet periods
  • CI/CD: Cloud Build, Spinnaker
  • Canary, blue/green deployments
• Basic SLIs—uptime alerts fired when APIs unresponsive
  • Improve with Cloud Monitoring

Business Requirements

• Expand to new locations—consistent experience
  • Global Load Balancing
  • Regional compute
  • Cloud CDN
• Support 10x as many concurrent users
  • Autoscaling: MIGs
• Collect user metrics—monetize product
  • BigQuery or Cloud Bigtable for time-series data
  • Cloud Logging, Cloud Domains
• Ensure regulatory compliance (e.g. GDPR)
• Reduce infrastructure management time and cost
  • IaC, e.g. Terraform
  • Migrate on-prem to Google Cloud
• Adopt Google best practices:
  • Workflows/process—application lifecycle management
    • Agile, CI/CD, canary deployments
  • Define SLIs and SLOs
    • Cloud Monitoring

Technical Requirements

• Secure communications between on-prem and Google Cloud
  • TLS, Cloud VPN, Cloud Interconnect
• Usage metrics/monitoring
  • BigQuery
• APIs—authentication/authorization
  • OpenAPI, Cloud Endpoints/Google Cloud API Gateway/Apigee
  • Internal apps: Identity Aware Proxy
• Implement faster/more accurate validation of new features
  • Canary deployments, A/B testing
• Logging/performance metrics—actionable information, debugging and alerts
  • Cloud Logging, Cloud Monitoring, Cloud Debugger
• Scale to meet user demands
  • MIGs, Google Cloud Load Balancing
  • Migrate to serverless: App Engine, Cloud Run, Cloud Functions

Potential Solution Design

References

• HipLocal case study overview