@nahisaho/yata-scale

YATA Knowledge Graph Scale-up Package - Distributed graph architecture, high-performance indexing, and multi-level caching for large-scale knowledge graphs.

Features

• Distributed Graph Architecture: Hash/Range/Graph-based partitioning with automatic rebalancing
• High-Performance Indexes: B+Tree, Full-text, Graph indexes with Bloom filters
• Multi-Level Cache: L1 (hot), L2 (warm), L3 (cold) with automatic promotion/demotion
• Parallel Query Processing: Worker pool, pipelined execution, async I/O
• Sync & Replication: Vector clock, conflict resolution, offline support

Installation

npm install @nahisaho/yata-scale

Quick Start

import { createYataScale } from '@nahisaho/yata-scale';

const scale = createYataScale({
  shards: [
    { id: 'shard-0', type: 'local', path: './data/shard-0' },
    { id: 'shard-1', type: 'local', path: './data/shard-1' },
  ],
  partitionStrategy: 'hash',
  cache: {
    l1Size: 1000,
    l2Size: 10000,
  },
});

await scale.initialize();

// Add entities (auto-sharding)
await scale.addEntity({
  type: 'class',
  name: 'UserService',
  namespace: 'app.services',
});

// Distributed query
const result = await scale.query({
  entityFilters: { types: ['class'] },
  textSearch: 'Service',
});

await scale.close();

Performance Targets

| Metric | Target | |--------|--------| | Single entity lookup | < 1ms p99 | | Complex graph queries | < 100ms p99 | | Read QPS | ≥ 10,000 | | Memory (1M entities) | < 4GB | | Max entities | 100M+ |

License

MIT