Stripe logo

Stripe has developed a new real-time streaming analytics system for its Billing service to enable businesses to quickly adapt to changing customer behavior by providing high-quality billing data with minimal latency. The system replaces traditional batch processing, which had a 24-hour lag, with an architecture that supports real-time subscription updates and allows queries to reflect data changes within 15 minutes. Key components of the system include the use of Apache Flink for real-time data updates and Apache Pinot's new query engine for flexible, low-latency data aggregation. This setup enables users to visualize metrics like monthly recurring revenue (MRR) in real time, while maintaining the ability to customize metric definitions without sacrificing data consistency. The updated system ensures that users experience a responsive Dashboard, with query latency less than 300 milliseconds, and allows historical recalculations to align with real-time updates even as metric definitions change. As Stripe continues to refine the system, it aims to further reduce data latency and enhance the Dashboard with more data dimensions and metrics.