Airbnb's observability engineering team has published details of a large-scale migration away from StatsD and a proprietary Veneur-based aggregation pipeline toward a modern, open-source metrics stack

InfoQ is a leading online platform for software developers, architects, and technical leaders, providing news, articles, presentations, and interviews on a wide range of topics, including agile practices, DevOps, microservices, and emerging technologies. With a focus on quality content and expert insights, InfoQ helps professionals stay informed about the latest trends, best practices, and industry developments. Developers can learn from real-world experiences, gain  knowledge, and connect with peers in the global software community through InfoQ's diverse and engaging content.

InfoQ

Airbnb's observability team migrated from a StatsD/Veneur-based metrics pipeline to one built on OpenTelemetry Protocol (OTLP), the OpenTelemetry Collector, and VictoriaMetrics' vmagent, now processing over 100 million samples per second. Key decisions included dual-emitting metrics during transition, switching high-cardinality services to delta temporality to reduce memory pressure, and selecting vmagent for its streaming aggregation and horizontal sharding. A significant correctness challenge emerged: Prometheus counter reset semantics caused systematic undercounting for low-frequency, high-cardinality metrics. The fix was a 'zero injection' technique in vmagent that emits a synthetic zero on first flush, implicitly initializing counters. The migration reduced costs by roughly an order of magnitude and the aggregation tier became a general-purpose transformation layer. The post also contextualizes similar migrations at Flipkart and Shopify, framing the StatsD exit and move to open-source Prometheus-anchored stacks as an industry-wide pattern.

Airbnb Migrates High-Volume Metrics Pipeline to OpenTelemetry