Introducing Grafana Alloy, A Distribution of the OTel Collector
summary
Grafana Alloy is a distribution of the OpenTelemetry Collector that actually merges OTel and Prometheus: components form programmable pipelines, secret exports read from Vault with live refresh, and Prometheus discovery scrapes targets from Kubernetes or a local kubelet. Clustering uses HTTP/2 gossip so clustering converges within a second and ownership recalculates on node changes, yielding roughly 50% cheaper than hashmod sharding for pull workloads. Configuration supports reusable blueprints that can be pulled from Git/S3/HTTP for GitOps updates and hot reloads.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 2:36Say hello to Grafana Alloy
- 4:32Alloy configuration syntax
- 5:13Alloy components
- 6:04Programmable pipeline architecture
- 8:25Native Kubernetes support
- 11:43OpenTelemetry phase (demo)
- 16:00How Alloy supports clustering
- 18:57Modules in Alloy
- 25:24Fleet management
related talks
21mInstant Measurement and Data Ingestion: Demo
17mFrom noise to signal: piping your way to clear observability
37mShould I Use OTel (collectors), or Is Prometheus Good Enough?
40mInstalling OpenTelemetry, Today and Tomorrow
50mOpen Telemetry: The Rise of Open Standards in Observability
36mShould I use OTel (collectors), or is Prometheus good enough?
30mWith OpenTelemetry, ComplyAdvantage overhauled its observability (twice)
23mImplementation of Unified Observability at Scale From Scratch