Uber has introduced uForwarder, a powerful Kafka consumer proxy, into the open-source community. Designed for efficiency in event-driven microservices architectures, uForwarder seamlessly manages the immense scale of Uber’s data processing needs, handling trillions of messages and petabytes daily across over a thousand downstream consumer services.
Replacing traditional direct Kafka consumer client implementations, uForwarder utilizes a gRPC-driven push interface, which simplifies consumer logic by centralizing offset management. This strategic design addresses various operational challenges while optimizing resource utilization.
Among its notable features is context-aware routing facilitated by Kafka message headers, allowing for effective workload isolation without necessitating topic proliferation. To further enhance processing efficiency, the system incorporates an out-of-order commit tracker, which redirects stalled messages to a dead letter queue, mitigating issues of head-of-line blocking.
uForwarder also boasts a sophisticated auto-rebalancing mechanism that dynamically redistributes partitions according to real-time metrics such as CPU usage, memory load, and throughput. Additionally, the DelayProcessManager offers granular partition-level control, enabling the pausing and resumption of message processing as needed.
Through these innovative features, Uber’s uForwarder significantly improves hardware utilization, reduces consumer lag, and effectively isolates workloads across different environments, making it a robust solution for large-scale event-driven and microservices architectures.

Collections

Uber has open-sourced uForwarder, a Kafka consumer proxy built to handle trillions of messages and petabytes of data daily across thousands of downstream services. It replaces direct Kafka consumer clients with a gRPC-based push interface that centralizes offset management. Key features include context-aware routing via Kafka message headers for workload isolation, an out-of-order commit tracker with dead letter queue support to prevent head-of-line blocking, auto-rebalancing based on real-time CPU/memory/throughput metrics, and a DelayProcessManager for partition-level pause/resume control. The result is improved hardware utilization, reduced consumer lag, and better workload isolation at massive scale.

Uforwarder: Uber's Scalable Kafka Consumer Proxy for Efficient Event-Driven Microservices