A specialized microservice handling FX rate aggregation from multiple liquidity providers, intelligent rate selection, and distribution to downstream services with sub-millisecond latency.
The client's existing rate infrastructure was a bottleneck—built as a monolithic component, it couldn't scale to meet growing demand. They needed a dedicated service that could aggregate rates from multiple providers, apply business logic, and serve thousands of concurrent requests without breaking a sweat.
We built the service in Go for maximum performance, with an in-memory rate cache updated via WebSocket connections to liquidity providers. The service implements circuit breakers for provider failover and publishes rate updates to downstream consumers via Redis pub/sub.
The service handles 10,000+ requests per second with p99 latency under 5ms. Provider failover is automatic and transparent to consumers. Rate freshness improved from 5-second updates to real-time streaming.
Go, GraphQL, Redis, Kubernetes, Prometheus, Grafana