Apache Solr

Mature open-source enterprise search platform built on Apache Lucene. Solr predates Elasticsearch and remains widely deployed; alongside Elasticsearch, OpenSearch, and Vespa it is one of the reference engines for engine-selection discussions in this vault.


Core Model

  • Core — a single physical index with its own configuration (analogous to a Vespa application, though a Vespa app can hold multiple indexes).
  • managed-schema — field and field-type definitions (indexed, stored, docValues, analyzers).
  • solrconfig.xml — request handlers, caching, update processors.
  • Update endpoint — document ingestion; query endpoint — Lucene/eDisMax queries.

Notable Features

  • BM25 relevance via Lucene (see BM25).
  • More Like This (MLT) — similarity search from a seed document; the classic lexical analog of dense-vector “more like this” (nearestNeighbor in Vespa; kNN in OpenSearch).
  • Function queries — score manipulation via functions/UDF-like expressions.
  • Faceting — mature faceted-search support (see Faceted Search).
  • Late-interaction reranking — recent ColBERT-on-Solr work; see ColBERT Comes to Apache Solr.

Position in the Ecosystem

Solr and Elasticsearch share the Lucene foundation and overlap heavily in capability; Solr is often favored for its configurability and community governance (Apache). It is a common source system in platform migrations to OpenSearch or Vespa — see Migration between Search Engines.

BM25 · Full-Text Search · Faceted Search · Late Interaction · ColBERT · Search Platforms

Articles