Elasticsearch
The distributed search and analytics engine. Store, search, and analyze data at any scale, with vector search, AI toolkits, and a high-performance retrieval engine.
Latest: Stack 9.4.1 (Mar 28, 2026) · Serverless deployed Apr 1, 2026
- semantic_text field type Simplify semantic search — no pipeline setup needed, just define the field and start querying
- ES|QL improvements New functions, cross-cluster support, and dense vector operations directly in ES|QL queries
- Inference API expansion Connect to external AI models (OpenAI, Cohere, Anthropic, Hugging Face, Azure, Google) for embeddings and reranking
- Breaking changes in 9.x Review removed settings, deprecated APIs, and behavioral changes before upgrading from 8.x
Install and deploy
-
Install and run Elasticsearch on your own infrastructure.
-
Run Elasticsearch on Elastic Cloud, Kubernetes, or ECE.
-
Production guidance, upgrades, cluster health, and diagnostics.
Index and ingest data
-
Pre-process documents before indexing with processors that parse, transform, and enrich your data.
-
Manage time-series data with a single resource that spans multiple backing indices.
-
Index, update, delete, and retrieve documents with the core document APIs.
Search and query
-
Write expressive queries using the JSON-based Domain Specific Language — from simple term filters to compound boolean queries.
-
A pipe-based query language purpose-built for filtering, transforming, and analyzing data — usable from the API, Kibana Discover, or directly in search.
-
Use EQL for event sequences, SQL for familiar syntax, or Painless for custom scoring and scripted queries.
Aggregations
-
Summarize and compute statistics over your data — counts, averages, histograms, date ranges, and more.
-
Compute single numeric values from field data — min, max, avg, sum, percentiles, stats, and more.
-
Group documents into buckets, then chain pipeline aggregations to compute on those buckets.
Data modeling
-
Define how documents and their fields are stored, indexed, and analyzed.
Common field types -
Control how text is tokenized and indexed to optimize relevance and search recall.
-
All Elasticsearch node settings — network, thread pools, index defaults, security, and more.
Manage data
-
Automate index policies to move data through hot, warm, cold, and frozen tiers — and eventually delete it.
-
Set built-in retention and rollover directly on data streams, without a separate ILM policy.
-
Back up indices and data streams to a remote repository and restore them when needed.
AI and vector search
-
Search by meaning rather than exact keywords using dense or sparse vector embeddings.
-
Find the nearest neighbors to a query vector for similarity-based retrieval and hybrid search.
-
Connect Elasticsearch to external AI model providers for embeddings, reranking, and completions.
Security
-
Encrypt communications between nodes and between clients and the cluster.
-
Verify user identity with native realms, LDAP, Active Directory, SAML, OIDC, PKI, and more.
-
Control what users and applications can access using roles, privileges, and field- and document-level security.
Clients and integrations
-
Official Elasticsearch clients for your preferred programming language.
-
High-level Python libraries that simplify building queries, mappings, and bulk operations.
-
Get data into Elasticsearch using Fleet, Logstash, Beats, or search connectors.
Reference
-
Conventions, common options, compatibility guarantees, and the full interactive API specification.
-
Write custom scripts in Painless for queries, aggregations, ingest processors, and runtime fields.
-
What's new, deprecated, and fixed in each Elasticsearch release.
-
Diagnose and fix common Elasticsearch cluster, performance, and operational issues.