쇼 HN: OpenTelemetry x DuckDB(Ducklake) x Clojure

hackernews | | 📦 오픈소스
#ai 모델 #ai 딜 #claude #s3 #xet #머신러닝/연구
원문 출처: hackernews · Genesis Park에서 요약 및 분석

요약

DuckDB와 SQLite를 기반으로 하는 오픈 소스 OpenTelemetry 백엔드가 공개되어, 단일 컨테이너로 로그, 메트릭, 트레이스의 수집부터 저장 및 쿼리까지 처리할 수 있습니다. OTLP를 통해 데이터를 수집하며, 텔레메트리의 새로운 필드를 자동으로 쿼리 가능한 열로 변환하는 기능을 제공합니다. 또한 내장된 쿼리 빌더와 워터폴 뷰를 통해 데이터를 시각화하고, Alertmanager와 호환되는 웹훅 알림으로 경고를 설정할 수 있습니다.

본문

Free, open-source OpenTelemetry backend powered by DuckDB and SQLite Logs, metrics, and traces all in a single process. - Full OpenTelemetry support. Ingests traces, logs, and metrics via OTLP/gRPC and OTLP/HTTP. - Single process. One container runs the entire stack: ingestion, storage, query, and UI. - Automatic schema evolution. New fields in your telemetry automatically become queryable columns. No manual schema management. - DuckDB + DuckLake. Telemetry stored as open Parquet files. Fast analytical queries, no heavy infrastructure. - Explore. Query and visualize traces, logs, and metrics with a built-in query builder and trace waterfall view. - Alert Rules. Define alert rules with Alertmanager-compatible webhook notifications. - Notebooks. Compose and share investigative workflows combining queries and notes. - Agent native. Built-in OAuth PKCE and an agent skill package let LLM agents query telemetry, manage alerts, and edit notebooks out of the box. Wide events over scattered logs. Traditional logging produces dozens of narrow, context-free lines per request that are optimized for writing, not querying. We believe in wide events: single, rich records per unit of work that carry all the context you need. User info, business data, feature flags, timings. O11yLite's automatic schema evolution and high-cardinality columnar storage are designed specifically to make this style of instrumentation practical. Send us your 50-field spans and we'll make every field queryable without configuration. Smaller but capable. The backend for observability doesn't need to be a distributed fleet of indexers and query nodes. Advances in embedded analytical databases like DuckDB mean a single process can handle what used to take a cluster. O11yLite leans into this: one container, one volume, minimal moving parts. Built to realize the potential of your 7000 MB/s NVMe drive. It sustains 140K events/s ingestion on a single M1 Max MacBook Pro and aims to serve teams from small startups through upper-mid-size organizations. Built for you and your agents. O11yLite offers first-class agent support through lean, solid primitives and a builtin agent skill. The agent skill is considered the first class interface just like UI. docker run -d \ -p 80:80 \ -v o11ylite-data:/data \ ghcr.io/o11ylite/o11ylite:latest Then: - Open http://localhost for the UI - Point your OTLP exporters at localhost:80 . The built-in proxy routes both gRPC and HTTP on the same port - Installation — Docker and Helm chart setup - Kubernetes OpenTelemetry Setup — Collect logs, traces, and metrics from your Kubernetes cluster - Authentication — OIDC setup, API keys, and scopes - Agent Native — First-class interface for LLM agents - FAQ — Frequently asked questions See DEVELOPMENT.md for setup instructions, architecture details, and contribution workflow. O11yLite is licensed under the GNU Affero General Public License v3.0. It's free for both personal and commercial use under AGPL copyleft terms. For commercial use without AGPL obligations, contact zhming0.

Genesis Park 편집팀이 AI를 활용하여 작성한 분석입니다. 원문은 출처 링크를 통해 확인할 수 있습니다.

공유

관련 저널 읽기

전체 보기 →