Skip to main content

Performance Examples

Benchmark and optimization examples for measuring and improving Grafio performance.

Run Benchmarks​

npm run perf

For accurate heap measurements:

npm run perf:gc

Customizing the Performance Test Suite​

The performance test suite is located at tests/perf/entryPoint.ts. You can customize it to test different scales and scenarios.

1. Modify Scale Configurations​

Edit the SCALES array to change test sizes:

const SCALES: ScaleConfig[] = [
  { label: 'Small  (10k nodes)',  nodeCount: 10_000,  edgesPerNode: 3 },
  { label: 'Medium (50k nodes)',  nodeCount: 50_000,  edgesPerNode: 3 },
  { label: 'Large  (100k nodes)', nodeCount: 100_000, edgesPerNode: 3 },
];

2. Choose Scenario Builder​

Use either buildCommonScenarios (for native API) or buildCommonScenariosCypher (for Cypher queries):

// For Cypher-based benchmarks
const scenarios = buildCommonScenariosCypher(scale.nodeCount, 0.1);

// For native API benchmarks
const scenarios = buildCommonScenarios(scale.nodeCount, 0.1);

3. Custom Scenario Structure​

Scenarios have this structure:

interface BenchmarkScenario {
  name: string;
  category: string;  // 'Write' | 'Read' | 'Navigation' | 'Traversal' | 'Aggregation'
  fn: (meta: GraphMeta) => Promise<unknown>;
}

4. Run Custom Benchmarks​

import { buildGraph, runScenario } from 'grafio/shared/testing/perf';

async function customBenchmark() {
  // Build graph with custom parameters
  const meta = await buildGraph(50_000, 3);
  
  // Create and run a custom scenario
  const scenario = {
    name: 'Custom Query',
    category: 'Traversal',
    fn: async (meta) => {
      // Your benchmark logic here
      return graph.traverse('start').depth(1..5).execute();
    }
  };
  
  const result = await runScenario(scenario, meta);
  console.log(`Mean: ${result.meanMs}ms, Ops/sec: ${result.opsPerSec}`);
}

Performance Tips​

  1. Use indexes — createIndex() for O(1) property lookups
  2. Enable caching — GraphManager.init() with appropriate cache config
  3. Limit results — use maxResults in traversal options
  4. Use Cypher LIMIT — for large result sets
  5. Use typed traversals — specify relationship types for faster queries

See the Caching Guide for cache configuration.