LoginSign Up
← Back to all sparks
M

Meilisearch

DEVOPS
Velocity6.3

Open-source, fast, typo-tolerant search engine API.

www.meilisearch.com ↗

Meilisearch grinds toward scale: sharded federated search and a Dynamic Search Rules revamp.

search-engineperformanceshardingfederated-searchopen-sourcebreaking-changes
◆Current state
Meilisearch is on a tight v1.4x-to-v1.50 cadence of real engineering: synonym-storage rewrites (up to 13x faster), CVE patches, and now v1.50's Dynamic Search Rules revamp plus federated document fetch across sharded deployments. The feed also carries prototype 'squash-me' tags that are pre-release profiling builds, not shipped versions.
◆Where it's heading
The arc points at large, multi-node deployments — sharding, federated search, and relevance rules that behave predictably under load — while hardening security. Experimental features like Dynamic Search Rules and the render-template route are maturing through breaking-change iterations before stabilizing.
◆Prediction
Expect v1.50's Dynamic Search Rules to keep churning through breaking changes toward a stable API, with continued work on federated and sharded search performance.

◆Recent moves

  1. 2d ago

    Prototype build: profile search-request speedups

    A pre-release prototype build for profiling search-request latency, not a shipped version — internal 'squash-me' measurement work.

    View source ↗
  2. 2d ago

    Prototype build: profile search-request speedups (0)

    A second pre-release prototype build in the same search-speedup profiling series; not a user-facing release.

    View source ↗
  3. 3d ago

    v1.50: Dynamic Search Rules revamp + sharded federated fetch

    v1.50 revamps Dynamic Search Rules with precedence semantics and adds federated document fetch for sharded setups — a real, if breaking, refinement of the scale features Meilisearch has been building toward.

    View source ↗
  4. 16d ago

    Prototype build: add missing OpenAPI route descriptions

    A prototype build adding missing OpenAPI route descriptions for docs; pre-release housekeeping, not a shipped change.

    View source ↗
  5. 17d ago

    v1.49: lazy synonym loading, up to 13x faster

    v1.49 makes synonyms load lazily, cutting the performance hit on high-synonym indexes by up to 13x — a concrete relevance-at-scale win consistent with the performance trajectory.

    View source ↗
  6. 24d ago

    v1.48.3: S3 snapshot race + federated dedup fixes

    A patch fixing an S3-snapshot race condition and a remote-federated-search duplicate-document bug; maintenance hardening, not new capability.

    View source ↗