Veneur is a distributed, fault-tolerant observability pipeline. It's fast, memory-efficient, and capable of aggregating global metrics.

It's designed to be:
Memory-efficient: Perform aggregation in constant space with HyperLogLog and t-digests.
Distributed: Run host-local instances, but also get global aggregate metrics.
Pluggable: Archive data in Datadog, LightStep, SignalFx, or a custom backend.
Parallelized: Scale metrics processing with a configurable worker pool.

To get started, check out the documentation. Learn how to deploy Veneur on Kubernetes or on bare metal. See how Veneur is being used at Stripe.