M.I.M.I.R - Multi-agent Intelligent Memory & Insight Repository

/** * Large-Scale Benchmark Suite for NornicDB vs Neo4j * * Datasets: * 1. Movies Dataset (~280 nodes, ~900 edges) - Classic Neo4j benchmark * 2. Social Network (~500 nodes, ~3500 edges) - Pokec-style * 3. E-commerce (~1000 nodes, ~1800 edges) - Products, reviews, users * * Total: ~1780 nodes, ~6200 relationships * * Compares performance between: * - NornicDB (drop-in replacement): bolt://localhost:7687 * - Neo4j: bolt://localhost:7688 * * Run with: npm run bench:large */ import { bench, describe, beforeAll, afterAll } from 'vitest'; import neo4j, { Driver, Session } from 'neo4j-driver'; // Configuration const NORNICDB_URI = process.env.NORNICDB_URI || 'bolt://localhost:7687'; const NEO4J_URI = process.env.NEO4J_URI || 'bolt://localhost:7688'; const NEO4J_USER = process.env.NEO4J_USER || 'neo4j'; const NEO4J_PASSWORD = process.env.NEO4J_PASSWORD || 'password'; let nornicdbDriver: Driver; let nornicdbSession: Session; let neo4jDriver: Driver; let neo4jSession: Session; // ============================================================================= // Data Generators // ============================================================================= const genres = ['Action', 'Comedy', 'Drama', 'Sci-Fi', 'Horror', 'Romance', 'Thriller', 'Documentary']; const firstNames = ['John', 'Jane', 'Michael', 'Sarah', 'David', 'Emma', 'Chris', 'Lisa', 'Tom', 'Kate']; const lastNames = ['Smith', 'Johnson', 'Williams', 'Brown', 'Jones', 'Garcia', 'Miller', 'Davis', 'Wilson', 'Taylor']; const cities = ['New York', 'Los Angeles', 'Chicago', 'Houston', 'Phoenix', 'Philadelphia', 'San Antonio', 'San Diego', 'Dallas', 'San Jose']; const categories = ['Electronics', 'Clothing', 'Books', 'Home', 'Sports', 'Toys', 'Beauty', 'Food']; const tiers = ['Bronze', 'Silver', 'Gold', 'Platinum']; // ============================================================================= // Dataset Loaders // ============================================================================= async function loadMoviesDataset(s: Session) { // Create 100 actors for (let i = 0; i < 100; i++) { await s.run( 'CREATE (a:Actor {name: $name, born: $born})', { name: `${firstNames[i % 10]} ${lastNames[Math.floor(i / 10)]}`, born: 1950 + (i % 50) } ); } // Create 30 directors for (let i = 0; i < 30; i++) { await s.run( 'CREATE (d:Director {name: $name, born: $born})', { name: `Director_${i}`, born: 1940 + (i % 40) } ); } // Create 150 movies for (let i = 0; i < 150; i++) { await s.run( 'CREATE (m:Movie {title: $title, released: $released, genre: $genre})', { title: `Movie_${i}`, released: 1980 + (i % 44), genre: genres[i % genres.length] } ); } // Create ACTED_IN relationships (5 actors per movie = 750 relationships) for (let movieId = 0; movieId < 150; movieId++) { for (let j = 0; j < 5; j++) { const actorIdx = (movieId * 5 + j) % 100; await s.run( `MATCH (a:Actor {name: $actorName}), (m:Movie {title: $movieTitle}) CREATE (a)-[:ACTED_IN {role: $role}]->(m)`, { actorName: `${firstNames[actorIdx % 10]} ${lastNames[Math.floor(actorIdx / 10)]}`, movieTitle: `Movie_${movieId}`, role: `Role_${j}` } ); } } // Create DIRECTED relationships (1 director per movie = 150 relationships) for (let i = 0; i < 150; i++) { await s.run( `MATCH (d:Director {name: $directorName}), (m:Movie {title: $movieTitle}) CREATE (d)-[:DIRECTED]->(m)`, { directorName: `Director_${i % 30}`, movieTitle: `Movie_${i}` } ); } } async function loadSocialNetwork(s: Session) { // Create 500 users for (let i = 0; i < 500; i++) { await s.run( 'CREATE (u:Person {id: $id, name: $name, age: $age, city: $city})', { id: i, name: `User_${i}`, age: 18 + (i % 60), city: cities[i % cities.length] } ); } // Create ~3500 FOLLOWS relationships (7 per user on average) for (let i = 0; i < 500; i++) { const followCount = 5 + (i % 5); // 5-9 follows per user for (let j = 0; j < followCount; j++) { const targetId = (i + j * 7 + 1) % 500; if (targetId !== i) { await s.run( `MATCH (a:Person {id: $from}), (b:Person {id: $to}) CREATE (a)-[:FOLLOWS {since: $since}]->(b)`, { from: i, to: targetId, since: `202${i % 4}-0${(j % 9) + 1}-01` } ); } } } } async function loadEcommerceData(s: Session) { // Create 300 products for (let i = 0; i < 300; i++) { await s.run( 'CREATE (p:Product {id: $id, name: $name, price: $price, stock: $stock, category: $category})', { id: i, name: `Product_${i}`, price: 10 + (i % 990), stock: i % 1000, category: categories[i % categories.length] } ); } // Create 200 customers for (let i = 0; i < 200; i++) { await s.run( 'CREATE (c:Customer {id: $id, name: $name, email: $email, tier: $tier})', { id: i, name: `Customer_${i}`, email: `customer${i}@test.com`, tier: tiers[i % tiers.length] } ); } // Create 500 orders with PLACED relationships for (let i = 0; i < 500; i++) { const customerId = i % 200; await s.run( `MATCH (c:Customer {id: $customerId}) CREATE (o:Order {id: $orderId, date: $date, total: $total}) CREATE (c)-[:PLACED]->(o)`, { customerId, orderId: i, date: `2023-${String((i % 12) + 1).padStart(2, '0')}-15`, total: 20 + (i % 480) } ); } // Create 800 reviews with REVIEWED relationships for (let i = 0; i < 800; i++) { const productId = i % 300; const customerId = i % 200; await s.run( `MATCH (p:Product {id: $productId}), (c:Customer {id: $customerId}) CREATE (c)-[:REVIEWED {rating: $rating}]->(p)`, { productId, customerId, rating: 1 + (i % 5) } ); } } async function loadAllDatasets(s: Session, dbName: string) { console.log(` → Loading Movies dataset into ${dbName}...`); await loadMoviesDataset(s); const movieCount = await s.run('MATCH (n) WHERE n:Actor OR n:Director OR n:Movie RETURN count(n) as c'); console.log(` ✓ Movies: ${movieCount.records[0].get('c')} nodes`); console.log(` → Loading Social Network into ${dbName}...`); await loadSocialNetwork(s); const socialCount = await s.run('MATCH (p:Person) RETURN count(p) as c'); console.log(` ✓ Social: ${socialCount.records[0].get('c')} users`); console.log(` → Loading E-commerce data into ${dbName}...`); await loadEcommerceData(s); const ecomCount = await s.run('MATCH (n) WHERE n:Product OR n:Customer OR n:Order RETURN count(n) as c'); console.log(` ✓ E-commerce: ${ecomCount.records[0].get('c')} nodes`); // Verify totals const countResult = await s.run('MATCH (n) RETURN count(n) as nodeCount'); const edgeResult = await s.run('MATCH ()-[r]->() RETURN count(r) as edgeCount'); console.log(` ✓ ${dbName} Total: ${countResult.records[0].get('nodeCount')} nodes, ${edgeResult.records[0].get('edgeCount')} relationships\n`); } // ============================================================================= // SETUP AND TEARDOWN (at module level - required for vitest bench) // ============================================================================= beforeAll(async () => { console.log('\n╔════════════════════════════════════════════════════════════════════╗'); console.log('║ NornicDB vs Neo4j - Large-Scale Benchmark Suite ║'); console.log('╚════════════════════════════════════════════════════════════════════╝\n'); // Connect to NornicDB console.log(`Connecting to NornicDB at ${NORNICDB_URI}...`); try { nornicdbDriver = neo4j.driver(NORNICDB_URI); nornicdbSession = nornicdbDriver.session(); await nornicdbSession.run('RETURN 1'); console.log('✓ Connected to NornicDB\n'); await nornicdbSession.run('MATCH (n) DETACH DELETE n'); console.log(' → Cleared existing NornicDB data'); await loadAllDatasets(nornicdbSession, 'NornicDB'); } catch (error) { console.error('✗ Failed to connect to NornicDB:', error); } // Connect to Neo4j console.log(`Connecting to Neo4j at ${NEO4J_URI}...`); try { neo4jDriver = neo4j.driver(NEO4J_URI, neo4j.auth.basic(NEO4J_USER, NEO4J_PASSWORD)); neo4jSession = neo4jDriver.session(); await neo4jSession.run('RETURN 1'); console.log('✓ Connected to Neo4j\n'); await neo4jSession.run('MATCH (n) DETACH DELETE n'); console.log(' → Cleared existing Neo4j data'); await loadAllDatasets(neo4jSession, 'Neo4j'); } catch (error) { console.error('✗ Failed to connect to Neo4j:', error); } console.log('─'.repeat(72) + '\n'); }, 600000); // 10 minute timeout for loading both DBs afterAll(async () => { console.log('\n' + '─'.repeat(72)); console.log('Cleaning up...'); if (nornicdbSession) { await nornicdbSession.run('MATCH (n) DETACH DELETE n').catch(() => {}); await nornicdbSession.close(); } if (nornicdbDriver) await nornicdbDriver.close(); if (neo4jSession) { await neo4jSession.run('MATCH (n) DETACH DELETE n').catch(() => {}); await neo4jSession.close(); } if (neo4jDriver) await neo4jDriver.close(); console.log('✓ Cleanup complete\n'); }, 30000); // ============================================================================= // NORNICDB BENCHMARKS // ============================================================================= describe('NornicDB Large-Scale', () => { // Movies queries bench('Movies: Count actors', async () => { await nornicdbSession.run('MATCH (a:Actor) RETURN count(a) as count'); }); bench('Movies: Find actors born after 1970', async () => { await nornicdbSession.run('MATCH (a:Actor) WHERE a.born > 1970 RETURN a.name, a.born'); }); bench('Movies: Movies by genre', async () => { await nornicdbSession.run('MATCH (m:Movie) WHERE m.genre = "Action" RETURN m.title, m.released'); }); bench('Movies: Actors with movie count', async () => { await nornicdbSession.run(` MATCH (a:Actor)-[:ACTED_IN]->(m:Movie) RETURN a.name, count(m) as movieCount ORDER BY movieCount DESC LIMIT 10 `); }); bench('Movies: Directors with avg movie year', async () => { await nornicdbSession.run(` MATCH (d:Director)-[:DIRECTED]->(m:Movie) RETURN d.name, avg(m.released) as avgYear, count(m) as movieCount `); }); bench('Movies: Co-actors (2-hop)', async () => { await nornicdbSession.run(` MATCH (a:Actor {name: 'John Smith'})-[:ACTED_IN]->(m:Movie)<-[:ACTED_IN]-(coactor:Actor) RETURN DISTINCT coactor.name LIMIT 20 `); }); // Social Network queries bench('Social: Count users by city', async () => { await nornicdbSession.run(` MATCH (p:Person) RETURN p.city, count(p) as userCount ORDER BY userCount DESC `); }); bench('Social: Average age by city', async () => { await nornicdbSession.run(` MATCH (p:Person) RETURN p.city, avg(p.age) as avgAge, min(p.age) as minAge, max(p.age) as maxAge `); }); bench('Social: Users with follower count', async () => { await nornicdbSession.run(` MATCH (p:Person)<-[:FOLLOWS]-(follower:Person) RETURN p.name, count(follower) as followers ORDER BY followers DESC LIMIT 10 `); }); bench('Social: Mutual follows', async () => { await nornicdbSession.run(` MATCH (a:Person)-[:FOLLOWS]->(b:Person)-[:FOLLOWS]->(a) RETURN a.name, b.name LIMIT 20 `); }); bench('Social: Friends of friends (2-hop)', async () => { await nornicdbSession.run(` MATCH (p:Person {id: 1})-[:FOLLOWS]->(friend)-[:FOLLOWS]->(fof) WHERE fof <> p RETURN DISTINCT fof.name LIMIT 20 `); }); // E-commerce queries bench('Ecommerce: Products by category', async () => { await nornicdbSession.run(` MATCH (p:Product) RETURN p.category, count(p) as productCount, avg(p.price) as avgPrice ORDER BY productCount DESC `); }); bench('Ecommerce: Top rated products', async () => { await nornicdbSession.run(` MATCH (c:Customer)-[r:REVIEWED]->(p:Product) RETURN p.name, avg(r.rating) as avgRating, count(r) as reviewCount ORDER BY avgRating DESC LIMIT 10 `); }); bench('Ecommerce: Customer order totals', async () => { await nornicdbSession.run(` MATCH (c:Customer)-[:PLACED]->(o:Order) RETURN c.name, c.tier, sum(o.total) as totalSpent, count(o) as orderCount ORDER BY totalSpent DESC LIMIT 10 `); }); bench('Ecommerce: Revenue by customer tier', async () => { await nornicdbSession.run(` MATCH (c:Customer)-[:PLACED]->(o:Order) RETURN c.tier, sum(o.total) as revenue, count(o) as orders, avg(o.total) as avgOrder ORDER BY revenue DESC `); }); bench('Ecommerce: Low stock expensive products', async () => { await nornicdbSession.run(` MATCH (p:Product) WHERE p.stock < 100 AND p.price > 500 RETURN p.name, p.price, p.stock, p.category ORDER BY p.stock ASC `); }); // Complex queries bench('Complex: Multi-aggregation', async () => { await nornicdbSession.run(` MATCH (p:Product) RETURN p.category, count(p) as products, sum(p.price) as totalValue, avg(p.price) as avgPrice, min(p.price) as minPrice, max(p.price) as maxPrice ORDER BY totalValue DESC `); }); bench('Complex: Large result set (500 rows)', async () => { await nornicdbSession.run(` MATCH (a:Actor)-[:ACTED_IN]->(m:Movie) RETURN a.name, m.title, m.released LIMIT 500 `); }); bench('Write: Create and delete node', async () => { await nornicdbSession.run(` CREATE (t:Temp {id: 99999, name: 'Benchmark Test'}) WITH t DELETE t `); }); bench('Write: Create and delete relationship', async () => { await nornicdbSession.run(` MATCH (a:Actor), (m:Movie) WITH a, m LIMIT 1 CREATE (a)-[r:TEMP_REL]->(m) DELETE r `); }); }); // ============================================================================= // NEO4J BENCHMARKS (same queries for comparison) // ============================================================================= describe('Neo4j Large-Scale', () => { // Movies queries bench('Movies: Count actors', async () => { await neo4jSession.run('MATCH (a:Actor) RETURN count(a) as count'); }); bench('Movies: Find actors born after 1970', async () => { await neo4jSession.run('MATCH (a:Actor) WHERE a.born > 1970 RETURN a.name, a.born'); }); bench('Movies: Movies by genre', async () => { await neo4jSession.run('MATCH (m:Movie) WHERE m.genre = "Action" RETURN m.title, m.released'); }); bench('Movies: Actors with movie count', async () => { await neo4jSession.run(` MATCH (a:Actor)-[:ACTED_IN]->(m:Movie) RETURN a.name, count(m) as movieCount ORDER BY movieCount DESC LIMIT 10 `); }); bench('Movies: Directors with avg movie year', async () => { await neo4jSession.run(` MATCH (d:Director)-[:DIRECTED]->(m:Movie) RETURN d.name, avg(m.released) as avgYear, count(m) as movieCount `); }); bench('Movies: Co-actors (2-hop)', async () => { await neo4jSession.run(` MATCH (a:Actor {name: 'John Smith'})-[:ACTED_IN]->(m:Movie)<-[:ACTED_IN]-(coactor:Actor) RETURN DISTINCT coactor.name LIMIT 20 `); }); // Social Network queries bench('Social: Count users by city', async () => { await neo4jSession.run(` MATCH (p:Person) RETURN p.city, count(p) as userCount ORDER BY userCount DESC `); }); bench('Social: Average age by city', async () => { await neo4jSession.run(` MATCH (p:Person) RETURN p.city, avg(p.age) as avgAge, min(p.age) as minAge, max(p.age) as maxAge `); }); bench('Social: Users with follower count', async () => { await neo4jSession.run(` MATCH (p:Person)<-[:FOLLOWS]-(follower:Person) RETURN p.name, count(follower) as followers ORDER BY followers DESC LIMIT 10 `); }); bench('Social: Mutual follows', async () => { await neo4jSession.run(` MATCH (a:Person)-[:FOLLOWS]->(b:Person)-[:FOLLOWS]->(a) RETURN a.name, b.name LIMIT 20 `); }); bench('Social: Friends of friends (2-hop)', async () => { await neo4jSession.run(` MATCH (p:Person {id: 1})-[:FOLLOWS]->(friend)-[:FOLLOWS]->(fof) WHERE fof <> p RETURN DISTINCT fof.name LIMIT 20 `); }); // E-commerce queries bench('Ecommerce: Products by category', async () => { await neo4jSession.run(` MATCH (p:Product) RETURN p.category, count(p) as productCount, avg(p.price) as avgPrice ORDER BY productCount DESC `); }); bench('Ecommerce: Top rated products', async () => { await neo4jSession.run(` MATCH (c:Customer)-[r:REVIEWED]->(p:Product) RETURN p.name, avg(r.rating) as avgRating, count(r) as reviewCount ORDER BY avgRating DESC LIMIT 10 `); }); bench('Ecommerce: Customer order totals', async () => { await neo4jSession.run(` MATCH (c:Customer)-[:PLACED]->(o:Order) RETURN c.name, c.tier, sum(o.total) as totalSpent, count(o) as orderCount ORDER BY totalSpent DESC LIMIT 10 `); }); bench('Ecommerce: Revenue by customer tier', async () => { await neo4jSession.run(` MATCH (c:Customer)-[:PLACED]->(o:Order) RETURN c.tier, sum(o.total) as revenue, count(o) as orders, avg(o.total) as avgOrder ORDER BY revenue DESC `); }); bench('Ecommerce: Low stock expensive products', async () => { await neo4jSession.run(` MATCH (p:Product) WHERE p.stock < 100 AND p.price > 500 RETURN p.name, p.price, p.stock, p.category ORDER BY p.stock ASC `); }); // Complex queries bench('Complex: Multi-aggregation', async () => { await neo4jSession.run(` MATCH (p:Product) RETURN p.category, count(p) as products, sum(p.price) as totalValue, avg(p.price) as avgPrice, min(p.price) as minPrice, max(p.price) as maxPrice ORDER BY totalValue DESC `); }); bench('Complex: Large result set (500 rows)', async () => { await neo4jSession.run(` MATCH (a:Actor)-[:ACTED_IN]->(m:Movie) RETURN a.name, m.title, m.released LIMIT 500 `); }); bench('Write: Create and delete node', async () => { await neo4jSession.run(` CREATE (t:Temp {id: 99999, name: 'Benchmark Test'}) WITH t DELETE t `); }); bench('Write: Create and delete relationship', async () => { await neo4jSession.run(` MATCH (a:Actor), (m:Movie) WITH a, m LIMIT 1 CREATE (a)-[r:TEMP_REL]->(m) DELETE r `); }); });