Presenters

Source

Scaling Citizen Payments: Lessons from Building a System for 20 Billion Transactions a Month 🚀

Ever wondered how a payment system can handle billions of transactions every month, reliably and securely? The recent presentation on the journey of UPI (Unified Payments Interface) and its underlying infrastructure provided a fascinating deep dive into the architecture, evolution, and key principles behind a system designed for citizen-scale impact. Let’s explore the key takeaways!

The Challenge: Payments for Everyone 🌐

The core goal is simple: to create a payment system accessible and reliable for every citizen. This isn’t just about handling numbers; it’s about building trust and ensuring inclusivity. The scale is staggering: currently processing 20 billion transactions monthly, with ambitious plans to reach 30 billion, and eventually 10 billion daily! But simply scaling existing systems isn’t enough. It requires a fundamental rethinking of architecture and a commitment to innovation.

From Monolith to Modular: The Architectural Evolution 🛠️

The journey hasn’t been linear. Here’s a look at the key milestones:

  • UPI 1.0: The Monolithic Foundation: The initial architecture started with a traditional monolithic design – a common starting point. While functional, it lacked the flexibility needed for rapid innovation.
  • Modular Architecture: Embracing Change: Recognizing the need for agility, the system evolved to a modular design, allowing for the addition of new product lines and features without disrupting the core functionality.
  • Layered Architecture (Future – 10 Bat): Looking ahead, a layered architecture using patterns like CQRS (Command Query Responsibility Segregation) will further enhance separation of concerns and improve performance. Imagine a system where read operations (like checking your balance) are handled independently from write operations (like making a payment) – that’s the power of CQRS!
  • Performance Boost: API Gateway Consolidation: Future plans involve consolidating the Web Application Firewall (WAF), Load Balancer (LB), and API Gateway into a single layer. This will dramatically reduce latency and improve overall performance.

Data at Scale: Databases, Caching, and Stream Processing 💾

Handling massive transaction volumes requires a robust data infrastructure. Here’s how the system has evolved:

  • Early Days: PostgreSQL for Back-Office: Initially, PostgreSQL was used for back-office operations.
  • Real-Time Data: Cassandra and Kafka: Transaction data is now pushed to Cassandra via Kafka, enabling real-time analytics and reporting.
  • Eliminating Database Calls: The ultimate goal? To eliminate database calls in the critical path of transactions using stream processing – a game-changer for speed and efficiency!
  • Caching Strategies: From an initial Active-Passive setup, caching has evolved to leverage Radius for local caching and KDB (Key-Value Database) for global caching. Future exploration of off-heap caching promises even greater performance gains.
  • HSM Evolution: Security is paramount. The system started with primarily hardware-based HSMs, moved to an Active-Active deployment, and is now exploring edge-scale, edge-deployed HSMs and kernel-level encryption.

Open Source: Building for the Community 👾

The organization is committed to giving back to the open-source community:

  • Falcon: Kubernetes Deployment Helper: Falcon, a Kubernetes-based deployment helper for Hyperledger Fabric networks, is a prime example. It simplifies deployment and management, making it easier for others to build on similar technologies.
  • More to Come: Expect to see more open-source contributions in the future – a testament to the belief that collaboration drives innovation.

Key Principles for Building Citizen-Scale Systems ✨

What can we learn from this incredible journey? Here are the core principles:

  • Citizen-Scale Design: Always prioritize accessibility and reliability for everyone.
  • Open Source First: Leverage open-source technologies whenever possible – it fosters collaboration and accelerates innovation.
  • Always On: Design for continuous availability. Downtime is not an option.
  • Active-Active Architecture: Embrace redundancy and automatic failover.
  • Strict Quality of Service: Enforce strict quality of service levels to ensure consistent performance.
  • Zero Trust & Security by Design: Integrate security into every aspect of the design – never compromise on safety.
  • Standardization: Advocate for and adhere to standardized specifications for interoperability.

Glossary (Quick Reference)

  • UPI: Unified Payments Interface – a real-time payment system in India.
  • CQRS: Command Query Responsibility Segregation – a pattern separating read and write operations.
  • HSM: Hardware Security Module – a secure hardware device for cryptographic keys.
  • Kafka: A distributed streaming platform for real-time data pipelines.
  • Kubernetes: A container orchestration system.

The journey of UPI and its underlying infrastructure is a powerful example of how innovative architecture, a commitment to open source, and a focus on citizen-scale impact can create truly transformative technologies. What lessons from this presentation resonate most with your work?

Appendix