Skip to content

Logchef vs Elasticsearch and Kibana

Logchef is not a drop-in replacement for Elasticsearch. It is a log UI that reads ClickHouse, VictoriaLogs or Grafana Loki, so using it as an ELK alternative means moving log storage as well as swapping Kibana. If you use Elasticsearch only to store logs and Kibana Discover to search them, that move gets you SQL, single sign-on, an audit log and alert notifications with no paid subscription. If you rely on Elastic for APM, SIEM, machine learning or ranked search, stay on Elastic.

Pick Logchef if your logs are in ClickHouse, VictoriaLogs or Loki (or you are willing to put them there) and you want search, dashboards, alerts and team access control from one self-hosted binary.

Pick Elasticsearch and Kibana if you need one platform for logs, metrics, traces, security analytics and application search, or if relevance-ranked full-text search matters to you.

OpenSearch and OpenSearch Dashboards sit on the Elasticsearch side of this comparison. They started as a fork of Elasticsearch and Kibana and use the same index-based storage model, so the architectural points below apply to them too. The licensing and paid-tier points do not.

Logchef Elasticsearch and Kibana
What it is Query UI and control plane over log storage you already run Search and analytics engine (Elasticsearch) with its own UI (Kibana)
License AGPLv3 Source available under AGPLv3, SSPL 1.0 or Elastic License 2.0, your choice. Files licensed solely under Elastic License 2.0 are found in x-pack; check each file header (LICENSE.txt)
Storage None of its own. Reads ClickHouse tables, VictoriaLogs and Loki Its own indices, on nodes you size and shard
Ingestion None. Your collectors write to the backend Owns it: Elastic Agent, Beats, Logstash, OpenTelemetry intake
Query languages LogchefQL, ClickHouse SQL, VictoriaLogs LogsQL, Loki LogQL KQL, Lucene syntax, ES|QL, Query DSL
Signals Logs only Logs, metrics, traces (APM), security events, search
Deployment One Go binary, SQLite or Postgres for metadata, plus your backend Elasticsearch cluster, Kibana, and shippers. Or Elastic Cloud
Auth and RBAC OIDC SSO, local auth, team-scoped source access and an audit log, all in the open-source release Native users, roles and Spaces are free. SSO, LDAP and audit logging need a paid subscription
Alerting Scheduled alerts with email, webhook, Slack and PagerDuty destinations Basic includes alerting rule types. Email, Slack, webhook and PagerDuty connectors need a paid subscription
Dashboards Built in: time series, stat, breakdown, table. Scheduled email reports Kibana dashboards and Lens
AI and MCP AI query generation with your own model key, built-in MCP server Elastic AI Assistant and Agent Builder (with MCP) are Enterprise tier
Pricing Free. You pay for the infrastructure you run Free Basic tier, paid self-managed subscriptions, or usage-based Elastic Cloud

Tier facts in this table come from Elastic’s subscriptions page as of November 2026.

A search index versus an analytical database

Section titled “A search index versus an analytical database”

Elasticsearch builds an inverted index for analyzed text and can rank matches by relevance, as its full-text search documentation explains.

Logchef leaves storage to ClickHouse, VictoriaLogs or Loki. ClickHouse is a columnar database: logs are rows, fields are typed columns, and aggregations are plain SQL. Text search in Logchef is substring and token matching, with no relevance score. For “show me errors from the payment service in the last hour, grouped by host”, that is enough. For “find the documents most similar to this phrase”, it is not.

Elastic has narrowed the storage gap for logs with its logsdb index mode, which is in the free Basic tier. One of its main space savings, synthetic _source, is an Enterprise feature from 8.17 onward, per the subscriptions page.

This is the difference most self-hosting teams hit first. In the free Basic tier, Elastic includes role-based access control, native and file-based users, Kibana Spaces, Discover, dashboards and alerting rules. It does not include:

  • Single sign-on. Elastic’s SAML docs say it directly: “Elastic Stack SSO is a subscription feature” (source). LDAP and Active Directory are paid too.
  • Alert notifications. Basic includes the Server log and Index connectors. Email, Slack, webhook and PagerDuty connectors are in the paid tiers.
  • Audit logging and field- and document-level security.

Logchef has no paid tier. OIDC single sign-on, team-scoped access to sources, service tokens, an audit log and email, webhook, Slack and PagerDuty alert destinations are in the open-source binary. Logchef grants access per source and has no field- or document-level security.

Elasticsearch and Kibana offered Apache 2.0 distributions until 2021. From version 7.11 the source moved to a dual SSPL and Elastic License model, and Elastic stopped shipping an Apache 2.0 distribution (licensing FAQ). In August 2024 Elastic announced AGPLv3 as a third option, so the source is open source again by the OSI definition. The paid features stay under the Elastic License.

OpenSearch took the other path. It was announced in January 2021 as a fork, is licensed under Apache 2.0, and is now a Linux Foundation project. Its security plugin supports SAML, OpenID Connect and LDAP without a subscription, so the paid-tier gap above is an Elastic point, not an OpenSearch one.

Kibana’s default language, KQL, is a filter language. Aggregations happen in Lens, or in ES|QL, Elastic’s newer piped language.

Logchef has the same split with different tools. LogchefQL is a short filter syntax (level="error" and status>=500) that also handles counts, sums and percentiles. When you need more, you switch to the backend’s native language: ClickHouse SQL with joins and window functions, LogsQL on VictoriaLogs, or LogQL on Loki. Logchef accepts a single ClickHouse SELECT, expands its time macros, and applies result limits. Queries that use Logchef macros need those placeholders expanded before use in clickhouse-client.

Elastic is a platform: Observability (logs, metrics, APM), Security (SIEM, endpoint) and Search are all built on the same cluster. Logchef does logs. It has no traces UI, no APM, no metrics store and no SIEM rules. It also does not ingest anything.

Logchef cannot query Elasticsearch or OpenSearch. The two run side by side, not stacked. A common split is to keep Elasticsearch for product search, APM or SIEM, and move high-volume application logs to ClickHouse or VictoriaLogs with Logchef on top.

If you want to keep Filebeat or Logstash, VictoriaLogs accepts the Elasticsearch bulk API, so an existing Elasticsearch output can often be pointed at it with a URL change. See Using VictoriaLogs with Logchef.

When Elasticsearch and Kibana is the better choice

Section titled “When Elasticsearch and Kibana is the better choice”
  • You need relevance-ranked search or vector search, not only log filtering.
  • You use Elastic APM, Elastic Security or ML anomaly detection and want logs correlated with them in one UI.
  • You want dynamic mapping to add fields automatically when you index documents.
  • You want index lifecycle management with hot, warm, cold, frozen, and delete phases. ILM is available in the versioned Elastic Stack, not Elasticsearch Serverless.
  • You already pay for an Elastic subscription or Elastic Cloud, and your team knows Kibana.
  • Your logs are already in ClickHouse, VictoriaLogs or Loki, or cluster cost is the reason you are leaving ELK.
  • You want SSO, team-scoped access, an audit log and alert notifications without a subscription.
  • You want SQL over logs, including joins against other ClickHouse tables.
  • You want a small query layer: one binary, TOML provisioning for teams and sources, and Prometheus metrics.
  • You want engineers and AI assistants to use the same access rules, through the CLI and the MCP server.

The Elasticsearch migration guide covers the whole path: re-pointing Filebeat, Logstash, Vector or the OpenTelemetry Collector, mapping ECS fields to ClickHouse columns, and a KQL to LogchefQL and SQL translation table. It also lists what you lose, so read that section before you commit.

For log analytics, yes, together with ClickHouse, VictoriaLogs or Loki as the store. For application search, APM or SIEM, no.

It replaces the parts of Kibana used for logs: Discover, the field sidebar, dashboards and alerting. It does not read Elasticsearch indices, so it is not a different front end for an existing Elastic cluster.

Can Logchef query Elasticsearch or OpenSearch?

Section titled “Can Logchef query Elasticsearch or OpenSearch?”

No. Logchef supports ClickHouse, VictoriaLogs and Grafana Loki sources.

The free portions of the source have offered AGPLv3 as an option since 2024, alongside SSPL and the Elastic License. The default distribution uses Elastic License 2.0. Features above the Basic tier still need a paid subscription.

No. LogchefQL covers the same filtering role with a similar feel. The migration guide has side-by-side examples.

Yes. Point them at VictoriaLogs through its Elasticsearch-compatible endpoint, or replace them with Vector or the OpenTelemetry Collector to write to ClickHouse.

Ready to try it? Start with the quickstart.