Introduction to Geode

Geode is an enterprise-ready graph database aligned with the ISO/IEC 39075:2024 Graph Query Language (GQL) 100% compliance. Built with Zig for performance and reliability, Geode provides QUIC+TLS transport, ACID transactions, and comprehensive enterprise security features.

What Geode Is

Geode is a production-ready graph database that combines:

  • ISO/IEC 39075:2024 compliance (see conformance profile)
  • Enterprise-grade security with authentication, encryption (TDE/FLE), and audit logging
  • QUIC+TLS transport for modern, secure, multiplexed connections
  • Built with Zig for memory safety and performance
  • Distributed query coordination supporting federated execution across shards

Core Differentiators

Standards Compliance (GQL)

Geode maintains 100% ISO/IEC 39075:2024 GQL compliance:

  • ✅ MATCH pattern syntax with variable-length paths (bounded expansions)
  • ✅ Aggregations (COUNT, SUM, AVG, MIN, MAX, GROUP BY)
  • ✅ Set operations (UNION, INTERSECT, EXCEPT)
  • ✅ Path helper functions (length(), nodes(), relationships())
  • ✅ Deterministic ordering and pagination policies (see conformance profile)

Why it matters: Standards alignment keeps queries portable and semantics predictable, with a clearly documented conformance scope.

Transaction Integrity

MVCC (Multi-Version Concurrency Control) + SSI (Serializable Snapshot Isolation):

  • True ACID semantics with serializable isolation
  • Optimistic concurrency control with conflict detection
  • Write-Ahead Log (WAL) for point-in-time recovery
  • Storage-level checksumming for data integrity
  • Session persistence across queries

Why it matters: Your graph data maintains consistency even under high concurrency, with the strongest isolation guarantees available.

Index Breadth

Geode provides specialized indexing for diverse workloads:

  • B-tree indexes: Range queries and sorting
  • Hash indexes: Fast equality lookups
  • Full-text search: BM25 ranking with tokenization
  • Spatial indexes: R-tree for geographic queries (LatLon, GeoPoint)
  • Vector similarity: HNSW (Hierarchical Navigable Small World) with SIMD acceleration
  • Specialized indexes: JSON path indexing, Patricia trie for network addresses, interval trees

Cost-Based Optimizer (CBO) and IndexOptimizer automatically select the best index for each query based on statistics and cardinality estimates.

Why it matters: No matter your query pattern—text search, geospatial, vector similarity, or graph traversal—Geode has a specialized index to accelerate it.

Security Stack

Enterprise security features built-in, not bolted-on:

  • Authentication: RBAC/ABAC with Argon2id password hashing, MFA support
  • Enhanced Row-Level Security (RLS): Policy-based access control (SELECT/INSERT/UPDATE/DELETE policies)
  • Transparent Data Encryption (TDE): Disk and WAL encryption with AES-256-GCM
  • Field-Level Encryption (FLE): Selective encryption with blind indexes for searchable encrypted data
  • Audit Logging: Tamper-evident logs with hash chaining, cryptographic signatures, and tracing IDs
  • KMS Integration: Multiple provider support (Vault, AWS KMS, environment-based)

Why it matters: Meet compliance requirements (GDPR, HIPAA, PCI) with battle-tested security controls, not custom implementations.

Distributed Query Coordination

Execute queries across federated graphs and sharded datasets:

  • Federated execution: Coordinate queries across multiple Geode instances
  • Result merging: Combine results from distributed shards with deterministic ordering
  • EXPLAIN FEDERATION: Visualize distributed query plans

Why it matters: Scale beyond a single node while maintaining GQL query semantics and transactional integrity.

Feature Highlights

If you care about…

Vector similarity search → Geode supports HNSW vector indexing with SIMD-accelerated distance calculations.

Full-text ranking → BM25 integration provides relevance scoring with optimizer awareness for efficient query planning.

Graph algorithms → Built-in support for centrality (PageRank, betweenness), community detection, shortest paths, and ML embeddings (Node2Vec, GraphSAGE, DeepWalk).

Spatial queries → Geographic types (LatLon, GeoPoint) with R-tree indexing for location-based queries.

Data types → 50+ specialized types including vectors (VectorF32/VectorI32), timestamps with timezone support, JSON/JSONB, binary data (Bytea), network addresses (IpAddr, Subnet), cryptographic hashes, and UUIDs.

Performance → Memory-mapped I/O, cost-based query optimization, and SIMD acceleration for hot paths.

Compliance and governance → 1,735 CANARY governance markers tracking 2,190+ requirements with evidence-based development methodology.

Architecture at a Glance

Geode’s architecture is designed for performance and reliability:

  • Storage Engine: Memory-mapped I/O with page-level checksums
  • Query Engine: Cost-based optimizer with index-aware rewriting
  • Transaction Manager: MVCC with write-ahead logging
  • Security Layer: Policy-driven RLS evaluation with encryption
  • Transport: QUIC+TLS for secure, multiplexed connections (no TCP fallback)

For architectural details, see the Architecture pages.

When to Choose Geode

Geode is ideal for:

  • OLTP graph workloads requiring strong isolation and ACID guarantees
  • Compliance-driven environments needing audit trails, encryption, and policy controls
  • Hybrid workloads combining graph traversal with vector similarity or full-text search
  • Standards-based applications leveraging portable GQL queries
  • Distributed systems requiring federated query coordination

Geode may not be the best fit for:

  • Simple key-value workloads (consider Redis or similar)
  • Pure analytics workloads without transactional requirements (consider graph analytics engines)
  • Extremely high-write scenarios requiring eventual consistency (Geode prioritizes consistency over availability)

Production Readiness

Current Status: Production Ready (v0.1.3+)

  • Test Coverage: 97.4% pass rate (1,644/1,688 integration tests)
  • Unit Tests: 100% pass rate (393/393 tests)
  • GQL conformance profile: see conformance profile
  • Governance: 1,735 CANARY markers tracking requirements with evidence

Architecture:

  • Memory-mapped I/O for efficient storage access
  • SIMD-accelerated vector distance calculations
  • Cost-based query optimization with statistics

Next Steps

Ready to get started?

Explore use cases:

Join the community: