ELK Stack
Elasticsearch + Logstash + Kibana — the open-source log management trio. Now also "Elastic Stack" with Beats.
Mindmap
The plain-English version
The ELK Stack is Elasticsearch (search engine + storage), Logstash (ingest pipeline), and Kibana (UI for querying and dashboards). With Beats (lightweight shippers like Filebeat), it becomes the Elastic Stack. The pre-eminent open-source log platform.
The problem it solves
For self-hosted log management at scale, ELK is the open standard. Search is fast, the UI is good, integrations exist for everything. Alternatives — Loki, OpenSearch (the AWS fork), Splunk, Datadog Logs — have their place but ELK is what most regulated, on-prem, or cost-sensitive teams settle on.
Alternatives
| Alternative | Type | When it wins |
|---|---|---|
| Sentry | errors | The error-tracking standard. Captures frontend and backend exceptions with full context. First tool teams add for production observability. |
| Datadog | APM | Comprehensive observability — metrics, logs, APM, RUM, all under one expensive roof. |
| Prometheus | metrics | The open-source metrics standard. Pull-based scraping, time-series database, the basis of most cloud-native observability. |
Deep links
The words you'll hear
- Elasticsearch
- Distributed search/analytics engine. JSON documents in indices.
- Logstash
- Ingest pipeline. Parses, transforms, routes. Heavyweight.
- Kibana
- Web UI for query, dashboards, alerting.
- Beats / Filebeat
- Lightweight shippers. Tail logs and ship them.
- Index / Shard
- Logical grouping / its physical pieces.
- ILM
- Index Lifecycle Management. Hot → warm → cold → delete based on age.
- OpenSearch
- AWS's open-source fork after Elastic re-licensed. Mostly compatible.
Bad vs. good prompt for ELK Stack
Why it works: Specifies single-node for dev (avoids the cluster setup pain), names the lifecycle policy explicitly, and asks for both files. Realistic dev setup, not the production-grade rabbit hole.
What bites real teams
Logs grow forever without ILM. Set retention before going to production.
Elasticsearch loves RAM. Undersized clusters fall over under search load. Plan capacity.
Many use cases don't need full Logstash; Filebeat → Elasticsearch directly is leaner.