Skip to content

Storage Model

Grafeo uses a hybrid storage model optimized for graph workloads.

Overview

graph TB
    subgraph "Logical Layer"
        NODES[Nodes]
        EDGES[Edges]
        PROPS[Properties]
    end

    subgraph "Physical Layer"
        COLS[Columnar Storage]
        ADJ[Adjacency Lists]
        IDX[Indexes]
    end

    subgraph "Persistence Layer"
        WAL[Write-Ahead Log]
        DATA[Data Files]
    end

    NODES --> COLS
    EDGES --> ADJ
    PROPS --> COLS
    COLS --> DATA
    ADJ --> DATA
    IDX --> DATA
    WAL --> DATA

Sections

  • Columnar Properties


    Type-specific columnar storage for properties.

  • Adjacency Lists


    Chunked adjacency lists for edge traversal.

  • Zone Maps


    Statistics for data skipping.

  • Compression


    Type-specific compression strategies.

  • Container Format


    .grafeo file format: copy-on-write checkpoints, a chained directory and checksummed chunks.

  • Ring Index


    Wavelet-tree based compact triple index for RDF data (~3x space reduction).