← all SaaS

Self-host Elastic Cloud

Managed search engine · Category: storage, queues & search license changed 2021

Elastic Cloud is the Elastic-managed hosting of Elasticsearch — full-text search, vector search, log analytics. Elastic's 2021 license change (SSPL/Elastic License v2) prompted the OpenSearch fork; the self-hostable replacements all do indexing + querying + relevance tuning, with vector search varying in maturity.

License change (2021). Apache-2.0 → SSPL/Elastic License v2 (Jan 2021); Elastic added an AGPL-3.0 option back in Aug 2024. AWS forked the last Apache release as OpenSearch in Apr 2021.

Elastic Cloud pricing anchor: Standard from ~$95/mo for a small cluster; Gold/Platinum/Enterprise tiers scale with hot-tier RAM.

OpenSearch opensearch-project/OpenSearch alive

GitHub
★ 12.9k · last commit today · 2789 open issues
License
Apache-2.0
AWS fork of Elasticsearch 7.10.2 (the last Apache-2.0 release); maintained by AWS + community.
Setup time
30min docker-compose (OpenSearch + OpenSearch Dashboards)
Monthly cost
$30-100/mo VPS — JVM heap drives memory needs; budget by data size + query load.
Migration sketch. Pull `opensearchproject/opensearch:2` and `opensearchproject/opensearch-dashboards:2` via the official docker-compose. Elasticsearch indices through 7.10.x are wire-compatible with OpenSearch — use the OpenSearch upgrade tool or `_reindex` from a snapshot. Above 7.10 (8.x), use cross-cluster `_reindex` from Elastic Cloud → OpenSearch over HTTPS. Kibana dashboards port to OpenSearch Dashboards with minor JSON edits.
Good fit forMost Elasticsearch workloads — OpenSearch is the closest one-for-one drop-in with the broadest ecosystem support.
Weak atSome Elasticsearch 8.x features (specific ML, vector ANN improvements) lag behind; check feature parity for your version.

Meilisearch meilisearch/meilisearch alive

GitHub
★ 57.5k · last commit 3d ago · 297 open issues
License
MIT
Setup time
5min single-binary or docker run
Monthly cost
$5-15/mo VPS — Rust binary is light; in-memory index sized to your data.
Migration sketch. Run `docker run -p 7700:7700 -e MEILI_MASTER_KEY='...' -v $HOME/meili-data:/meili_data getmeili/meilisearch:v1.10`. Migrate documents via Meili's `/indexes/{id}/documents` POST; transform Elasticsearch _source JSON to flat documents. Search is typo-tolerant out of the box, sub-100ms with no tuning. Vector search via `embedders` config with OpenAI/HuggingFace/Ollama embeddings.
Good fit forSite-search, in-app search, e-commerce search — the 'search box on a website' use case where Elasticsearch is overkill.
Weak atNot a full analytics engine — no aggregations across millions of documents the way Elasticsearch handles them.

Typesense typesense/typesense alive

GitHub
★ 25.8k · last commit 1d ago · 800 open issues
License
GPL-3.0
Setup time
10min single-binary or docker run
Monthly cost
$5-15/mo VPS.
Migration sketch. Pull `typesense/typesense:0.27.0` and run with `--api-key`. Typesense ships with an Elasticsearch-shaped HTTP API (close enough that some Elasticsearch clients work with adapters). Migrate via bulk `POST /collections/{name}/documents/import` with Elasticsearch _source. Vector search and federated search built-in.
Good fit forTeams that want the typo-tolerance of Algolia with an HTTP API that feels familiar to Elasticsearch users.
Weak atCluster operations are still maturing vs Elasticsearch/OpenSearch; sharding model less flexible.

In a terminal? npx os-alt elasticsearch-cloud prints this table — how the CLI works →