Skip to main content
Pinega
Use dark theme Menu
Research and engineering programme Database systems · Pinega Labs

Correctness under concurrency.

Pinega turns database-systems research into high-performance software across storage, transactions, optimisation, verification, and distributed execution. Pinega Engine is the first active implementation programme; it is not yet a production release.

Current programme status

Available now

Research and web platform

Versioned research artefacts, Strata design tokens, semantic web pages, and reproducible validation gates.

Active programme

Pinega Engine

An extension-oriented PostgreSQL OLTP architecture under design and implementation validation.

Portfolio direction

Optimisation, verification, and distributed systems

Research directions with explicit product potential, not shipped products or performance claims.

Programme thesis

Performance matters only when the result remains correct and explainable.

Pinega connects formal reasoning, systems experiments, production engineering, and commercial intent. The objective is reusable intellectual property and measurable customer value, not isolated demonstrations.

Correctness under concurrency

Histories, visibility, publication, lifetime protection, durability, and failure boundaries are explicit design objects.

Performance with evidence

Cache behaviour, NUMA placement, SIMD, storage media, and query planning are measured rather than advertised in advance.

Research to product

Papers and hypotheses move through models, verification, implementation, benchmarks, and documented product boundaries.

Technology programmes

One programme, several paths to defensible database software.

Programme labels describe active research and engineering boundaries. They do not imply commercial availability.

Active design

Pinega Engine

Version-first OLTP storage, a Pinega-owned shared buffer pool, PostgreSQL integration, and one PostgreSQL WAL boundary.

Inspect the engine programme
Research

Query optimisation and AI

Cardinality reasoning, query rewrites, MCTS/RL join ordering, and low-latency inference for future optimiser products.

Inspect the optimisation direction
Research platform

Correctness and verification

Executable models, deterministic testing, semantic diagrams, and tools for validating concurrent and transactional systems.

Inspect the verification direction

Featured active programme

Pinega Engine: an explicit OLTP path inside PostgreSQL.

The current architecture separates PostgreSQL integration, mutable versions, shared-frame lifetimes, and durability. It is an accepted design direction under implementation validation, not an installable release.

  1. 01
    PostgreSQL integrationTable AM callbacks, snapshots, transactions, and SQL-visible behaviour
    target
  2. 02
    Mutable version storeOut-of-place tuples, stable handles, and explicit visibility metadata
    design
  3. 03
    Shared buffer poolStable frames, protect/pin/revalidate, cache-aware placement, and reclamation
    research
  4. 04
    PostgreSQL WALOne write-ahead log and one transaction durability boundary
    decision
Proposed Pinega Engine OLTP path. Follow the Technology page for programme scope, decisions, and maturity.

Research-to-product discipline

Claims advance only when their evidence advances.

Source and formalise

Start from papers, standards, source code, tests, and measured system behaviour. Keep external evidence separate from inference.

Prototype and falsify

Turn architectural ideas into executable models, focused prototypes, deterministic schedules, and benchmarks with explicit failure criteria.

Implement and productise

Promote only validated capability into maintained software, documented support boundaries, licensing, and customer-facing claims.

Maturity ledger

What exists, what is being validated, and what remains proposed.

AreaCurrent stateEvidence boundary
YDMP research workspaceAvailableVersioned notes, MODEL/VERIFY artefacts, build commands, and CI
Pinega Strata and Web platformAvailableCanonical tokens, Web Components, static pages, and browser/accessibility gates
Scientific Diagram LanguageAvailableCanonical models, Web/Typst renderers, layout profiles, and review artifacts
Pinega Engine architectureUnder validationAccepted decisions and executable specimens; no production engine binary
Future product programmesResearchNo commercial availability or measured product claims yet

Start with the boundary

Understand the programmes, then inspect the evidence.

Technology describes what Pinega is building; Research explains why; Documentation shows what can be reproduced today.