Skip to main content

Scalability

Target​

100K customers, 1M+ transactions, high concurrent payments.

Strategies​

  • Horizontal pod autoscaling (HPA)
  • Kafka partitioning (12-24 partitions)
  • Database read replicas
  • Redis caching
  • Connection pool tuning
  • Async processing for non-critical paths
  • Backpressure via Kafka consumer pause

Bottleneck Analysis​

ComponentBottleneckMitigation
Payment ServiceDB writesPartition, connection pools
Ledger ServiceBalance updatesOptimistic locking, partition
KafkaPartition countIncrease partitions
GatewayConnection countALB scaling