What's New in Elasticsearch 8.0 – Key Features and Changes

Elasticsearch is a Lucene‑based distributed search engine with a multi‑tenant HTTP API and schema‑free JSON documents, available in many language clients. After nearly three years, version 8.0 was released with several notable enhancements.

7.x REST API Compatibility

Elasticsearch 8.0 introduces optional compatibility headers that allow 7.x‑compatible requests and responses, easing migration while still encouraging developers to adopt native 8.0 APIs.

Security Enabled by Default

Security features (authentication, authorization, TLS) are now enabled and configured automatically on first start, generating a registration token for Kibana or other nodes without manual certificate handling.

Known Issue – Password and Token Generation

On ARM or macOS M1 installations from archives, the initial elastic user password and Kibana token are not generated. Use the following commands to create them:

bin/elasticsearch-reset-password -u elastic

bin/elasticsearch-create-enrollment-token -s kibana

Improved System Index Protection

Direct access to system indices now requires the allow_restricted_indices privilege, and the built‑in elasticsuperuser role no longer has write access, encouraging management via Kibana or official APIs.

New KNN Search API (Technical Preview)

The KNN API leverages the dense_vector field to perform approximate nearest‑neighbor searches, offering faster performance on large datasets compared to the exact _score ‑based method.

Storage Savings for Keyword, match_only_text, and Text Fields

Updated inverted index encoding reduces storage for these fields (e.g., a 14.4% reduction for match_only_text messages, 3.5% overall disk savings).

Faster Indexing for geo_point, geo_shape, and Range Fields

Multi‑dimensional point indexing speed improves by 10‑15% according to Lucene benchmarks.

PyTorch Model Support for NLP

Elasticsearch now accepts externally trained PyTorch models for inference, expanding native NLP capabilities.

Other Changes

Aggregations : removed adjacency matrix settings, MovingAverage pipeline, deprecated _time and _term sorting, and date‑histogram interval deprecations.

Allocation : removed include_relocations setting.

Analysis : cleaned up versioned deprecations and removed pre‑configured delimited_payload_filter.

Authentication : added file and native realms by default, enforced NameID format handling, and ordered realm configuration.

Cluster Coordination : removed connection timeout handling and delayed state‑recovery support.

Distributed : removed sync‑flush and cluster.remote.connect settings.

Engine : rejected requests with both only_expunge_deletes and max_num_segments, removed per‑type index stats, and dropped translog retention settings.

Features/CAT APIs : deprecated local parameter for _cat/indices and _cat/shards.

Features/ILM+SLM : defaulted cluster.routing.allocation.enforce_default_tier_preference to true.

Features/Indices APIs : defaulted prefer_v2_templates to true, removed deprecated _upgrade API, eliminated include_type_name from REST layer, and removed template field from index templates.

Infra/Core : removed nodes/0 prefix from data paths, dropped bootstrap.system_call_filter, node.max_local_storage_nodes, Joda dependencies, and camel‑case date/time format names.

Packaging : removed SysV init support, Java HOME handling, and now requires Java 17 to run.

For full details, see the official Elastic blog post linked at the end of the article.