Elasticsearch is open source, again

Original Article ↗ Hacker News Discussion ↗

Licensing change & motivations

  • Elasticsearch is now dual-licensed (AGPL + Elastic License). Many welcome “free software again,” but argue the blog post spins the prior license change as a success instead of a mistake.
  • Several believe the real driver is competitive pressure: OpenSearch momentum, alternative search/vector DBs (Meilisearch, Typesense, ClickHouse, Loki, etc.), and weakened growth/stock performance.
  • Some say the change reduces FUD around “not really open source”; others note the Elastic License and CLA remain, so control still concentrates with Elastic.

Trust and willingness to switch back

  • Many commenters say “too late”: they’ve migrated to OpenSearch and see no reason to return, citing cost, effort, and fear of another future license flip.
  • Others appreciate the move and argue companies should be acknowledged when they revert to a free/open license.
  • Broken trust and the continued CLA are recurring reasons given for sticking with OpenSearch or other stacks.

Elasticsearch vs OpenSearch: adoption and features

  • Adoption/mindshare is contested: some anecdata say “everyone” now starts with OpenSearch; others cite DB-Engines and GitHub metrics showing Elasticsearch still far ahead.
  • Technically, for core document search both are very similar (same Lucene base). Differences:
    • Elasticsearch: richer observability/SIEM tooling, some powerful ingest processors, more mature docs and ecosystem, claims better vector-search performance.
    • OpenSearch: Apache 2.0, strong free-tier security (RBAC, SSO, alerts), tight AWS integration; criticized for weaker docs, flaky behavior, and rough Spark tooling.
  • New deployments reportedly lean toward OpenSearch, especially in security/logging and on AWS; many legacy ES users haven’t churned yet.

AWS, trademarks, and forks

  • Past “Amazon Elasticsearch Service” branding is widely seen as confusing or harmful to Elastic; Elastic sued on trademark and reached a settlement.
  • Commenters argue Elastic’s license change successfully forced AWS to maintain a clearly named fork (OpenSearch) and invest in it.
  • With AGPL, if AWS ever used upstream ES again and modified it, they’d need to publish changes; meanwhile Elastic can legally reuse Apache‑2.0 OpenSearch code, but not vice versa.

AGPL vs other licenses

  • AGPL is viewed by some engineers as clear and reasonable (publish your modifications when you offer the software as a network service).
  • In practice, many corporate legal departments ban AGPL due to perceived “viral” risks and lack of case law, making ES adoption harder than with Apache 2.0.
  • Elastic License, SSPL, BSL, and similar “source-available” licenses are debated:
    • Pro: protect vendors from hyperscalers reselling without contributing.
    • Con: not OSI‑approved, ambiguous terms (“competing service”, “substantial features”), and unfriendly to downstream hosting and community forks.

Open source, hyperscalers, and fairness

  • Strong disagreement over whether AWS and peers “exploit” permissive OSS or legitimately follow its rules.
  • Some argue permissive licensing plus cloud power depresses vendor business models and pushes relicensing; others say open source is inherently about allowing such use.
  • Broader philosophical debate: four freedoms vs “fair source,” whether licenses should explicitly block megacorp monetization, and whether strict copyleft (AGPL) is “ideal in the age of big tech.”

Miscellaneous

  • Elastic’s website, pricing clarity, and enterprise-heavy sales motions are heavily criticized.
  • The blog’s Kendrick Lamar track-title section headers are widely seen as distracting, cringey, or “LLM-ish.”