Brand Reliability through Highly Available Platforms
Digital workloads on active-active multi-region architecture
Global Automobile OEM
Challenge
- Brand dot com and loyalty management applications underwent planned and unplanned outages.
- Lift and shift migrated workloads on cloud was not scalable.
- Stateful applications suffered from persistence challenges of local cache.
- Service response time affective web page load and rendering time.
- High server upkeep cost during off peak hours.
Solution
- Redesigned application workloads into stateless microservices.
- Applications rewritten to work on serverless lambda based runtime with pay for use (on AWS).
- Designed highly available active active multi region infrastructure with latency and health based routing at DNS.
- Data persisted in Global No SQL AWS DynamoDB or MySQL Aurora Global DB.
Technologies
Benefits
- Brand.com and associated workloads benefitted with zero outages even during regional cloud service outages.
- Planned deployment outages were down to zero using blue green deployments.
- Infra cost brought down from 100 thousands to thousands per month.
- Campaigns, promotions, recalls and sales enjoyed interruption free services even during heavy traffic spikes.