Jeevan Henry Dsouza's team blog
AWS Lambda

AWS Lambda

Jeevan Henry Dsouza's photo
Jeevan Henry Dsouza
·

4 min read

Table of contents

Introduction

AWS Lambda is a serverless compute service provided by Amazon Web Services (AWS). It enables you to run code without provisioning or managing servers. With Lambda, you can execute your code in response to events and automatically manage the compute resources needed for that code. Lambda is a key component of serverless computing, allowing you to focus on writing code without the need to worry about server management.

Key features and characteristics of AWS Lambda include:

  1. Event-Driven Programming:

    • Lambda is designed for event-driven programming. It allows you to associate your code (function) with various AWS services or custom events, and the code is automatically triggered when those events occur.

  2. Serverless Architecture:

    • Lambda follows a serverless computing model. You don't need to provision or manage servers, and you only pay for the compute time consumed by your function.

  3. Wide Range of Triggers:

    • Lambda functions can be triggered by various events, including changes to data in an Amazon S3 bucket, updates to a DynamoDB table, incoming HTTP requests through API Gateway, modifications to an Amazon Kinesis stream, or custom events using CloudWatch Events.

  4. Programming Language Support:

    • Lambda supports multiple programming languages, including Node.js, Python, Java, Ruby, Go, .NET Core, and custom runtime environments.

  5. Auto Scaling:

    • Lambda automatically scales your application in response to incoming traffic. Each function execution is isolated, and AWS Lambda can run thousands of concurrent functions.

  6. Built-In Fault Tolerance:

    • AWS Lambda provides built-in fault tolerance. If a function execution fails, Lambda will automatically retry the execution. If the error persists, Lambda can send an event to a Dead Letter Queue (DLQ) for further analysis.

  7. Integrated Security:

    • Lambda functions can be associated with an execution role that defines the permissions needed to access other AWS services. IAM roles can be fine-tuned to control access to specific resources.

  8. Stateless Execution:

    • Lambda functions are designed to be stateless. They can access temporary storage (such as /tmp) during execution, but they don't retain state between invocations. For persistent storage, you can use other AWS services like Amazon S3 or DynamoDB.

  9. Cost Model:

    • With Lambda, you pay only for the compute time your code consumes. There are no charges when your code is not running. The service is billed in increments of 100 milliseconds.

AWS Lambda Cost

The pricing for AWS Lambda is based on two main factors: the number of requests made to your function and the time your code executes. As of my knowledge cutoff date in January 2022, the pricing model includes the following components:

  1. Requests:

    • You are charged based on the number of requests made to your function. This includes both invocations and the number of events processed by your function.

  2. Execution Time:

    • You are charged based on the time your code executes, rounded up to the nearest 100 milliseconds. The execution time is measured from the time your code begins executing until it returns or otherwise terminates.

Additional Considerations:

  • Free Tier:

    • AWS Lambda offers a free tier that includes 1 million free requests per month and 400,000 GB-seconds of compute time per month. This free tier is available to all AWS customers.

  • Concurrency:

    • If your function is invoked concurrently multiple times, each concurrent invocation is billed separately. The number of concurrent executions is a factor in determining the number of requests.

  • Memory Allocation:

    • The amount of memory allocated to your function affects both the price per request and the execution time. Higher memory allocations result in higher pricing.

  • Cold Starts:

    • The first time a Lambda function is invoked, it may experience a "cold start" where additional time is needed to initialize resources. Subsequent invocations benefit from reuse of the initialized resources.

  • Data Transfer:

    • Data transfer out from Lambda to other AWS services or the internet is subject to additional charges.

Example Pricing Scenario:

As of my last knowledge update, here's a simplified example to illustrate Lambda pricing:

Let's say you have a Lambda function that is invoked 1 million times in a month, with an average execution time of 100 milliseconds and a memory allocation of 128 MB.

  • Number of Requests: 1,000,000 requests (within the free tier)

  • Execution Time: 1,000,000 * 0.1 seconds = 100,000 seconds

  • GB-seconds: 100,000 seconds * (128 MB / 1024 MB) = 12,800 GB-seconds

lambda