Kotak Mahindra Bank Modernizes Microledgers Using Amazon DynamoDB for Predictable Low Latency at Scale

This article describes how Kotak Mahindra Bank modernized its core banking systems using AWS services to handle India's explosive UPI payment growth, which reached 18 billion transactions in March 2025.

• Built Real-time Transaction Service (RTS) using Amazon DynamoDB for distributed ledger functionality
• Implemented Read-Only Object Store (ROS) to offload balance and transaction inquiries from core system
• Increased monthly credit transactions from 237 million to 350 million post-implementation
• Reduced p95 latency for credits from 600ms to ~30ms using DynamoDB's predictable performance
• Dropped transaction failure rate from 1.5% to less than 0.01%
• Achieved 99.99% availability with p90 latency of 2.05ms for read operations
• Used Amazon EKS for microservices orchestration, Amazon MSK for data streaming, and ElastiCache for rate limiting
• System auto-switches between asynchronous and synchronous modes based on lag thresholds

Kotak Mahindra Bank successfully scaled its banking infrastructure to handle seasonal spikes and 50% annual UPI growth while maintaining core system availability for branch operations and regulatory reporting.