Key New Features and Changes in Elasticsearch 8.0 Release

Elasticsearch is a distributed, multi‑tenant full‑text search engine built on the Lucene library, offering an HTTP web interface and schema‑free JSON documents. Official clients exist for Java, .NET (C#), PHP, Python, Groovy, Ruby, and many other languages.

7.x REST API Compatibility

Version 8.0 adds optional compatibility headers that allow sending 7.x‑compatible requests to an 8.0 cluster and receiving 7.x‑compatible responses, easing migration.

Security Features Enabled by Default

From 8.0 onward, security (authentication, authorization, TLS) is enabled and configured automatically on first start. A registration token is generated for connecting Kibana or adding nodes without manual certificate handling.

Known Issues

On Linux ARM or macOS M1 installations from archives, the initial start does not generate the elastic user password or Kibana enrollment token. Use bin/elasticsearch-reset-password -u elastic to set the password, then bin/elasticsearch-create-enrollment-token -s kibana to create a token.

System Index Protection

System indices store internal configuration and data. Direct access is now blocked unless the allow_restricted_indices setting is set to true. The built‑in superuser role no longer has write privileges on system indices.

New KNN Search API

A technical preview of a KNN search API uses the dense_vector field to find the k nearest vectors, supporting recommendation engines and NLP‑based relevance ranking. The preview offers faster approximate searches compared with the previous exact script_score approach.

Storage‑Saving Field Mappings

Updates to the inverted index encoding reduce storage for keyword, match_only_text, and text fields. Benchmarks on a message field show a 14.4% reduction in index size and a 3.5% overall disk‑space saving.

Faster Indexing for Geo Types

Indexing speed for multi‑dimensional points used by geo_point, geo_shape, and range fields is improved by 10‑15% in Lucene‑level benchmarks.

PyTorch Model Support for NLP

Elasticsearch now accepts externally trained PyTorch models for inference, bringing modern NLP capabilities to the Elastic Stack.

Other Changes

Numerous deprecations and removals across aggregations, allocation, analysis, authentication, cluster coordination, distributed settings, engine options, CAT APIs, indices APIs, and core infrastructure. Highlights include removal of the MovingAverage pipeline aggregation, deprecation of _time and _term sorting, and the requirement of Java 17 to run Elasticsearch.

For full details see the official Elastic blog post: https://www.elastic.co/cn/blog/whats-new-elastic-8-0-0 .