repository · locator & rights
Introduction; installation; indexing; filtering
Link and original paraphrase only; upstream rights retained.
Composition / PRODUCT DOSSIER
P+PostgreSQL + pgvector
Vector retrieval alongside relational application data.
Self-managed · Open source · PostgreSQL and pgvector licensesRevision 2026-09-14.1 · unreviewedNeeds recheck · recheck 2026-10-140 attestations · checked 2026-09-14
What we do not know yet · 2 open gaps
- Which pinned versions, topology and service tier fit the intended workload?
- What are measured correctness, latency, recovery and total operating cost under that workload?
Applies to: Unpinned upstream documentation accessed 2026-09-14; pin engine, extension and service versions before use. · PostgreSQL plus an installed and enabled pgvector extension; operator manages both.
Applicability, 1 recheck triggers & verification plan
Applicability
Unpinned upstream documentation accessed 2026-09-14; pin engine, extension and service versions before use.
PostgreSQL plus an installed and enabled pgvector extension; operator manages both.
Excludes: Runtime verification; Workload benchmarks and cost comparisons; Unlisted extensions, service tiers and deployment configurations
Recheck when
- Upstream version, license, API, service plan or storage topology changes.
Needs recheck · due
Verification plan
- Pin the target configuration and define workload fixtures, access and budget.
- With authorization, test concurrent writes and recovery against explicit correctness expectations.
- For vector workloads, measure filtered recall against an exact-search baseline and record index settings.
Provenance & review
Testedknowhow database research
Commissioned by a Nimbalyst maintainer. Documentation synthesis; no independent review or product execution.
Revision 2026-09-14.1 · unreviewed · checked
0 attestations · Content: Unlicensed prototype; redistribution decision pending
Relational / Vector · Composition
PostgreSQL plus an installed and enabled pgvector extension; operator manages both.
Catalog components: PostgreSQL ↗
Target sources: pgvector README ↗ PostgreSQL overview ↗
Capability claims · 5 / 17 sourced
Standing condition: these are documentation claims at the source capture date, with outcome “claimed”; no runtime execution is recorded. Dossier applicability and limitations apply throughout. Expand a row for its statement, specific scope and contradictions. Numbered citations resolve to the source locator and capture date below. Missing evidence means Unknown, not No. Sources are unpinned unless marked “Pinned”; this marks source stability, not stronger support or runtime verification. Reuse counts refer to claims in the full dossier; repeated citations do not represent independent sources.
Yes Documented within scope Partial Limited form of the feature No Explicit evidence of absence Unknown No evidence recorded. These are support assessments, not quality scores or runtime verification.
Data model
SQL queriesYes
Explore SQL queries →
Uses Postgres SQL and joins.
Search
Vector similarity searchYes
Explore Vector similarity search →
Supports exact vector-distance search.
Approximate nearest neighborsYes
Explore Approximate nearest neighbors →
Provides HNSW and IVFFlat approximate indexes.
Metadata filteringYes
Explore Metadata filtering →
Combines vector queries with SQL WHERE filtering.
Scope: Approximate-index filtering can reduce returned results; tune scan settings and verify recall.
Consistency
Multi-record transactionsYes
Explore Multi-record transactions →
Retains Postgres transactional semantics.
Unknown · 12 capabilities without evidence
Interface images & local attachments
Sources · 2
vendor-documentation · locator & rights
Data types; transactions; recovery; security
Link and original paraphrase only; upstream rights retained.