DaLi / Coca-Cola · Large-scale infrastructure

Infrastructure for 500M+ coupons — DaLi / Coca-Cola, GEPP, Bimbo

Zero downtime during active campaigns. Redemptions processed in < 200ms even at peaks of 50k req/s.

500M+ coupon campaigns
Duration
6 months
Team
6 engineers
Industry
Beverages / Retail / FMCG
Stack
AWS · Node.js · Python · RDS Aurora · Load Balancers · Redis
Problem

Running campaigns of 500M+ free mobile data coupons for Coca-Cola, GEPP and Bimbo requires infrastructure that withstands traffic peaks of millions of simultaneous redemptions without failing.

Solution

We designed a distributed architecture on AWS with load balancers, multi-AZ RDS Aurora, Redis for validation caching and queues for asynchronous processing. Observability with CloudWatch + Grafana.

Implementation

How we built it

  1. 01Multi-region AWS architecture with automatic failover (us-east-1 + us-west-2)
  2. 02Multi-AZ RDS Aurora MySQL with read replicas to absorb validation queries
  3. 03Redis Cluster to cache valid coupons — cuts database load by 95% during peaks
  4. 04SQS queues + Lambda for asynchronous redemption processing (eventual consistency acceptable)
  5. 05API Gateway + Lambda edge for sub-200 ms validation anywhere in Mexico
  6. 06Real-time Grafana dashboards for brand operations teams
Measurable results

The impact in numbers

Campaign uptime
98.1%99.99%
+1.9pp
Average latency
1.8s180ms
-90%
Coupons processed
10M500M+
50×
Trend · ms
Latency before (ms)Latency after (ms)
01,2652,530

Zero downtime during active campaigns. Redemptions processed in < 200ms even at peaks of 50k req/s.

Have a similar project?

Tell us about your challenge and we will propose how we would solve it. No commitment.