LoginSign Up
← Back to all sparks
T

TypeDB

DEVOPS
Velocity5.0

Strongly typed polymorphic database with a native reasoning engine

typedb.com ↗

TypeDB ships a batch query primitive and schema annotation system, targeting production-scale deployments.

graph-databasequery-languagedistributedschema-evolutionopen-sourceperformance
◆Current state
TypeDB is in active feature development across the 3.12.x–3.13.x range, shipping minor versions every few weeks. The major capability addition this period is the `given` stage in 3.12.0—a batch parameterized query primitive that lets a single query run over multiple input rows, eliminating network round-trips, preventing TypeQL injection, and skipping repeated compilation overhead. Alongside that, 3.12.0 added `@doc` and `@meta` schema annotations and exposed RocksDB memory controls for production tuning. Subsequent releases have focused on schema management (type renaming in 3.12.2) and memory reliability (commit eviction, eager key cleanup in 3.13.0).
◆Where it's heading
TypeDB is building toward production-scale distributed deployments: clustered database import/export, UUID assignment for distributed user creation, and per-component memory limits are all infrastructure features that matter at scale. The three-stage query cache (parse, translate, compile) and the `given` stage reduce the per-query overhead that made TypeDB feel expensive at volume. Schema evolution tooling (type renaming, doc/meta annotations) is maturing, suggesting teams running TypeDB in production can now make schema changes without teardown.
◆Prediction
The clustering foundation being laid (import/export, UUIDs, metrics extensions API) is likely ahead of a TypeDB Cluster stability release or formal cluster documentation. The `given` stage will probably appear prominently in driver README updates and benchmarks next.

◆Recent moves

  1. 8d ago

    TypeDB 3.13.0

    TypeDB 3.13.0 adds commit eviction in the isolation manager (freeing memory from old commits still being read), an optional eager cleanup pass for deleted storage keys, and clustered import/export support for TypeDB Cluster deployments. Together, these reduce memory pressure and unblock data migration in distributed setups—follow-through from the cluster infrastructure work in 3.12.0.

    View source ↗
  2. 18d ago

    TypeDB 3.13.0-rc0

    Release candidate for 3.13.0, containing the same features. RC tags are internal validation milestones; no additional capability over what landed in the stable 3.13.0 a week later.

    View source ↗
  3. 1mo ago

    TypeDB 3.12.3

    3.12.3 fixes a class of database import rejections caused by inherited constraint combinations—independent sub-attributes and ownership/role specializations could trigger false validation failures during schema relaxation. A targeted correctness fix for teams doing schema evolution on inherited type hierarchies.

    View source ↗
  4. 1mo ago

    TypeDB 3.12.2

    3.12.2 ships type renaming (via `redefine old-label label new-label`) and a three-stage query cache (parse, translation, compile) that separates syntactic parsing from schema-dependent translation. Type renaming closes a long-standing schema evolution gap; the cache architecture makes repeated parameterized queries significantly faster by avoiding re-translation on each call.

    View source ↗
  5. 2mo ago

    TypeDB 3.12.1

    3.12.1 fixes a deadlock triggered by commits exceeding the RocksDB write buffer manager limit and corrects a string attribute retrieval bug where comparison bounds discarded more results than expected. Both are reliability patches for large-write workloads.

    View source ↗
  6. 2mo ago

    TypeDB 3.12.0

    ⚡ SPARK

    TypeDB 3.12.0 introduces the `given` stage—a new query primitive enabling batch parameterized queries that eliminate network round-trips, query-compilation overhead, and TypeQL injection risk. Combined with `@doc`/`@meta` schema annotations and configurable RocksDB memory limits, this is the most capability-dense release in recent history.

    View source ↗