Imported from jeremylongshore/tons-of-skills-marketplace (plugins/saas-packs/salesforce-pack/skills/salesforce-load-scale/SKILL.md). Install upstream with npx skills add jeremylongshore/tons-of-skills-marketplace --skill salesforce-load-scale. Copyright stays with the author (MIT).

Salesforce Load & Scale

Overview

Load testing, scaling strategies, and capacity planning for Salesforce integrations. Focus on API limit budgeting, Bulk API throughput, and handling Salesforce's unique constraint: org-wide shared limits.

Prerequisites

• k6 or Artillery load testing tool
• Sandbox or Developer org for testing (never load test production)
• Understanding of your org's API limit allocation
• Monitoring configured (see salesforce-observability)

Instructions

Step 1: Calculate API Limit Budget

const conn = await getConnection();
const limits = await conn.request('/services/data/v59.0/limits/');

const budget = {
  dailyMax: limits.DailyApiRequests.Max,
  currentlyUsed: limits.DailyApiRequests.Max - limits.DailyApiRequests.Remaining,
  remaining: limits.DailyApiRequests.Remaining,
  // Budget allocation
  integrationA: Math.floor(limits.DailyApiRequests.Max * 0.40), // 40% for primary sync
  integrationB: Math.floor(limits.DailyApiRequests.Max * 0.20), // 20% for secondary
  salesUsers: Math.floor(limits.DailyApiRequests.Max * 0.30),   // 30% for Salesforce UI users
  headroom: Math.floor(limits.DailyApiRequests.Max * 0.10),     // 10% buffer
};

console.table(budget);
// Example (Enterprise, 50 users): 150,000 daily calls
// Integration A: 60,000 | Integration B: 30,000 | Users: 45,000 | Buffer: 15,000

Step 2: Load Test with k6 (against Sandbox)

// salesforce-load-test.js
import http from 'k6/http';
import { check, sleep } from 'k6';

const SF_TOKEN = __ENV.SF_ACCESS_TOKEN;
const SF_INSTANCE = __ENV.SF_INSTANCE_URL;

export const options = {
  stages: [
    { duration: '1m', target: 5 },    // Ramp up
    { duration: '3m', target: 5 },    // Steady state
    { duration: '1m', target: 20 },   // Peak load
    { duration: '3m', target: 20 },   // Sustained peak
    { duration: '1m', target: 0 },    // Ramp down
  ],
  thresholds: {
    http_req_duration: ['p(95)<3000'],  // SF API calls are slower than typical SaaS
    http_req_failed: ['rate<0.01'],
  },
};

export default function () {
  // SOQL query
  const queryRes = http.get(
    `${SF_INSTANCE}/services/data/v59.0/query/?q=SELECT+Id,Name+FROM+Account+LIMIT+10`,
    { headers: { Authorization: `Bearer ${SF_TOKEN}` } }
  );
  check(queryRes, { 'query 200': (r) => r.status === 200 });

  // sObject retrieve
  const retrieveRes = http.get(
    `${SF_INSTANCE}/services/data/v59.0/sobjects/Account/describe`,
    { headers: { Authorization: `Bearer ${SF_TOKEN}` } }
  );
  check(retrieveRes, { 'describe 200': (r) => r.status === 200 });

  // Check rate limit headers
  const limitInfo = queryRes.headers['Sforce-Limit-Info'];
  if (limitInfo) {
    const [used, max] = limitInfo.replace('api-usage=', '').split('/');
    if (parseInt(used) / parseInt(max) > 0.8) {
      console.warn(`API usage at ${used}/${max}`);
    }
  }

  sleep(1); // Respect rate limits
}

# Get access token for load test
SF_ACCESS_TOKEN=$(sf org display --target-org my-sandbox --json | jq -r '.result.accessToken')
SF_INSTANCE_URL=$(sf org display --target-org my-sandbox --json | jq -r '.result.instanceUrl')

# Run load test (ONLY against sandbox)
k6 run \
  --env SF_ACCESS_TOKEN=$SF_ACCESS_TOKEN \
  --env SF_INSTANCE_URL=$SF_INSTANCE_URL \
  salesforce-load-test.js

Step 3: Bulk API Throughput Testing

// Bulk API 2.0 can process millions of records per job
// Key limits:
// - 15,000 Bulk API jobs/day
// - 150,000,000 records per 24hr rolling period
// - 10 concurrent Bulk API jobs

// Generate test data
function generateTestCsv(count: number): string {
  const lines = ['FirstName,LastName,Email,External_ID__c'];
  for (let i = 0; i < count; i++) {
    lines.push(`Test${i},User${i},test${i}@loadtest.example.com,LOAD-${i}`);
  }
  return lines.join('\n');
}

// Measure Bulk API throughput
const startTime = Date.now();
const results = await conn.bulk2.loadAndWaitForResults({
  object: 'Contact',
  operation: 'upsert',
  externalIdFieldName: 'External_ID__c',
  input: generateTestCsv(50000),
  pollInterval: 5000,
});

const duration = (Date.now() - startTime) / 1000;
const throughput = results.successfulResults.length / duration;

console.log({
  records: results.successfulResults.length,
  failures: results.failedResults.length,
  durationSeconds: duration.toFixed(1),
  recordsPerSecond: throughput.toFixed(1),
});
// Typical: 1,000-5,000 records/second depending on triggers and validation rules

Step 4: Scaling Strategies

// Strategy 1: Use Bulk API for large datasets (separate limit pool)
// Regular API: shared daily limit
// Bulk API: 15,000 jobs/day, unlimited records per job

// Strategy 2: Batch with sObject Collections (200 records/call)
// 100,000 API calls * 200 records/call = 20M records/day via REST

// Strategy 3: Reduce describe/metadata calls (cache aggressively)
// A single describe call can return 500+ fields — cache for hours

// Strategy 4: Use Composite API (25 operations per call)
// Replaces 25 individual calls with 1

// Strategy 5: Off-peak scheduling
// Run bulk jobs during business hours when sales users are active
// This ensures API limit usage is spread across the day

Step 5: Capacity Planning Table

Output

• API limit budget allocated across integrations
• Load test script targeting sandbox
• Bulk API throughput benchmarked
• Scaling strategies documented
• Capacity planning table for production

Error Handling

Examples

Benchmark Bulk API capacity without stressing production

Run a load test only in a designated sandbox with anonymized or generated records, a single test owner, and an agreed request and concurrency ceiling. Measure API consumption, Bulk API throughput, trigger execution time, and error rate in successive bounded rounds. Stop on the defined governor-limit or error threshold, preserve aggregate measurements, and return sandbox automation settings to their normal state before publishing the capacity recommendation.

Resources

• API Limits by Edition
• Bulk API 2.0 Limits
• k6 Documentation

Next Steps

For reliability patterns, see salesforce-reliability-patterns.