Skip to main content
This guide shows how to test serverless functions and measure cold start performance. Perfect for testing serverless APIs and function-based architectures.

​
Use Cases

  • Test serverless function response times
  • Measure cold start performance
  • Validate function scaling behavior
  • Test different serverless providers

​
Simple Implementation

from locust import task, HttpUser
import time
import random

class ServerlessTestUser(HttpUser):
    def on_start(self):
        # Serverless function endpoints to test
        self.functions = {
            "aws_lambda": {
                "url": "https://your-api-id.execute-api.region.amazonaws.com/prod/function",
                "name": "AWS Lambda"
            },
            "vercel": {
                "url": "https://your-app.vercel.app/api/function",
                "name": "Vercel Function"
            },
            "netlify": {
                "url": "https://your-app.netlify.app/.netlify/functions/function",
                "name": "Netlify Function"
            }
        }
        
        # Test payloads
        self.test_payloads = [
            {"message": "Hello World"},
            {"data": {"user_id": 123, "action": "test"}},
            {"query": "simple test query"},
            {}  # Empty payload
        ]

    @task(3)
    def test_aws_lambda(self):
        """Test AWS Lambda function"""
        if "aws_lambda" in self.functions:
            function = self.functions["aws_lambda"]
            payload = random.choice(self.test_payloads)
            
            start_time = time.time()
            
            with self.client.post(
                function["url"],
                json=payload,
                name="AWS Lambda"
            ) as response:
                response_time = (time.time() - start_time) * 1000
                
                if response.status_code == 200:
                    print(f"AWS Lambda: {response_time:.0f}ms")
                    
                    # Check for cold start indicators
                    if response_time > 1000:  # > 1 second might be cold start
                        print(f"AWS Lambda: Possible cold start ({response_time:.0f}ms)")
                else:
                    response.failure(f"AWS Lambda error: {response.status_code}")

    @task(3)
    def test_vercel_function(self):
        """Test Vercel Function"""
        if "vercel" in self.functions:
            function = self.functions["vercel"]
            payload = random.choice(self.test_payloads)
            
            start_time = time.time()
            
            with self.client.post(
                function["url"],
                json=payload,
                name="Vercel Function"
            ) as response:
                response_time = (time.time() - start_time) * 1000
                
                if response.status_code == 200:
                    print(f"Vercel Function: {response_time:.0f}ms")
                    
                    # Vercel functions typically have faster cold starts
                    if response_time > 500:
                        print(f"Vercel Function: Possible cold start ({response_time:.0f}ms)")
                else:
                    response.failure(f"Vercel Function error: {response.status_code}")

    @task(2)
    def test_netlify_function(self):
        """Test Netlify Function"""
        if "netlify" in self.functions:
            function = self.functions["netlify"]
            payload = random.choice(self.test_payloads)
            
            start_time = time.time()
            
            with self.client.post(
                function["url"],
                json=payload,
                name="Netlify Function"
            ) as response:
                response_time = (time.time() - start_time) * 1000
                
                if response.status_code == 200:
                    print(f"Netlify Function: {response_time:.0f}ms")
                    
                    if response_time > 800:
                        print(f"Netlify Function: Possible cold start ({response_time:.0f}ms)")
                else:
                    response.failure(f"Netlify Function error: {response.status_code}")

    @task(2)
    def test_cold_start_behavior(self):
        """Test cold start behavior by waiting between requests"""
        function_key = random.choice(list(self.functions.keys()))
        function = self.functions[function_key]
        
        # Wait to increase chance of cold start
        wait_time = random.uniform(30, 120)  # 30-120 seconds
        print(f"Waiting {wait_time:.0f}s to test cold start for {function['name']}")
        time.sleep(wait_time)
        
        start_time = time.time()
        
        with self.client.get(
            function["url"],
            name=f"Cold Start - {function['name']}"
        ) as response:
            response_time = (time.time() - start_time) * 1000
            
            if response.status_code == 200:
                print(f"Cold start test {function['name']}: {response_time:.0f}ms")
                
                # Analyze cold start performance
                if response_time > 2000:
                    print(f"Cold start {function['name']}: SLOW ({response_time:.0f}ms)")
                elif response_time > 1000:
                    print(f"Cold start {function['name']}: MODERATE ({response_time:.0f}ms)")
                else:
                    print(f"Cold start {function['name']}: FAST ({response_time:.0f}ms)")

    @task(1)
    def test_function_scaling(self):
        """Test how functions handle concurrent requests"""
        function_key = random.choice(list(self.functions.keys()))
        function = self.functions[function_key]
        
        # Make multiple rapid requests to test scaling
        concurrent_requests = 3
        results = []
        
        for i in range(concurrent_requests):
            start_time = time.time()
            
            with self.client.get(
                function["url"],
                name=f"Scaling Test - {function['name']}"
            ) as response:
                response_time = (time.time() - start_time) * 1000
                results.append(response_time)
                
                if response.status_code != 200:
                    response.failure(f"Scaling test failed: {response.status_code}")
        
        if results:
            avg_time = sum(results) / len(results)
            max_time = max(results)
            print(f"Scaling test {function['name']}: avg {avg_time:.0f}ms, max {max_time:.0f}ms")

    @task(1)
    def test_function_with_different_payloads(self):
        """Test functions with different payload sizes"""
        function_key = random.choice(list(self.functions.keys()))
        function = self.functions[function_key]
        
        # Test with different payload sizes
        payloads = [
            {"size": "small", "data": "x" * 100},
            {"size": "medium", "data": "x" * 1000},
            {"size": "large", "data": "x" * 10000}
        ]
        
        for payload in payloads:
            start_time = time.time()
            
            with self.client.post(
                function["url"],
                json=payload,
                name=f"Payload Test - {function['name']}"
            ) as response:
                response_time = (time.time() - start_time) * 1000
                
                if response.status_code == 200:
                    print(f"Payload test {function['name']} ({payload['size']}): {response_time:.0f}ms")
                else:
                    print(f"Payload test {function['name']} ({payload['size']}): ERROR {response.status_code}")

    @task(1)
    def test_function_error_handling(self):
        """Test how functions handle errors"""
        function_key = random.choice(list(self.functions.keys()))
        function = self.functions[function_key]
        
        # Test with invalid data to trigger errors
        error_payloads = [
            {"invalid": "data", "trigger_error": True},
            {"malformed": "json"},
            None  # No payload
        ]
        
        error_payload = random.choice(error_payloads)
        
        with self.client.post(
            function["url"],
            json=error_payload,
            name=f"Error Test - {function['name']}"
        ) as response:
            if response.status_code >= 400:
                print(f"Error test {function['name']}: Properly returned {response.status_code}")
            elif response.status_code == 200:
                print(f"Error test {function['name']}: Handled error gracefully")
            else:
                print(f"Error test {function['name']}: Unexpected response {response.status_code}")

​
Setup Instructions

  1. AWS Lambda: Deploy a test function and get the API Gateway URL
  2. Vercel: Deploy a function to Vercel and get the function URL
  3. Netlify: Deploy a function to Netlify and get the function URL
  4. Replace the placeholder URLs with your actual function endpoints

​
What This Tests

  • Response Times: Basic function execution speed
  • Cold Starts: Performance when functions haven’t run recently
  • Scaling: How functions handle concurrent requests
  • Error Handling: Function behavior with invalid inputs

​
Cold Start Analysis

Typical cold start times:
  • AWS Lambda: 100ms - 2000ms (depends on runtime and size)
  • Vercel Functions: 50ms - 500ms (generally faster)
  • Netlify Functions: 100ms - 800ms (moderate performance)

​
Performance Tips

  • Keep Functions Small: Smaller functions have faster cold starts
  • Use Appropriate Runtimes: Node.js typically faster than Python
  • Warm Functions: Regular requests keep functions warm
  • Monitor Memory Usage: Right-size function memory allocation

​
Common Issues

  • Cold Start Delays: First request after idle period is slow
  • Timeout Errors: Functions exceeding execution time limits
  • Memory Limits: Functions running out of allocated memory
  • Concurrent Limits: Too many simultaneous executions
⌘I