AWS Lambda FAQs

The code you run on AWS Lambda is uploaded as a “Lambda function”. Each function has associated configuration information, such as its name, description, entry point, and resource requirements. The code must be written in a “stateless” style i.e. it should assume there is no affinity to the underlying compute infrastructure. Local file system access, child processes, and similar artifacts may not extend beyond the lifetime of the request, and any persistent state should be stored in Amazon S3, Amazon DynamoDB, Amazon EFS, or another Internet-available storage service. Lambda functions can include libraries, even native ones.

To improve performance, AWS Lambda may choose to retain an instance of your function and reuse it to serve a subsequent request, rather than creating a new copy. To learn more about how Lambda reuses function instances, visit our documentation. Your code should not assume that this will always happen.

You can configure each Lambda function with its own ephemeral storage between 512MB and 10,240MB, in 1MB increments. The ephemeral storage is available in each function’s /tmp directory.

Each function has access to 512MB of storage at no additional cost. When configuring your functions with more than 512MB of ephemeral storage, you will be charged based on the amount of storage you configure, and how long your function runs, metered in 1ms increments. For comparison, in the US East (Ohio) region, the AWS Fargate ephemeral storage price is $0.000111 per GB-hour, or $0.08 per GB-month. Amazon EBS gp3 storage volume pricing in US East (Ohio) is $0.08 per GB-month. AWS Lambda ephemeral storage pricing is $0.0000000309 per GB-second, or $0.000111 per GB-hour and $0.08 per GB-month. To learn more, see AWS Lambda Pricing.

You can configure each Lambda function with its own ephemeral storage between 512MB and 10,240MB, in 1MB increments by using the AWS Lambda console, AWS Lambda API, or AWS CloudFormation template during function creation or update.

Yes. All data stored in ephemeral storage is encrypted at rest with a key managed by AWS.

You can use AWS CloudWatch Lambda Insight metrics to monitor your ephemeral storage usage. To learn more, see the AWS CloudWatch Lambda Insights documentation.

If your application needs durable, persistent storage, consider using Amazon S3 or Amazon EFS. If your application requires storing data needed by code in a single function invocation, consider using AWS Lambda ephemeral storage as a transient cache. To learn more, please see Choosing between AWS Lambda data storage options in web apps.

Yes. However, if you application needs persistent storage, consider using Amazon EFS or Amazon S3. When you enable Provisioned Concurrency for your function, your function's initialization code runs during allocation and every few hours, as running instances of your function are recycled. You can see the initialization time in logs and traces after an instance processes a request. However, initialization is billed even if the instance never processes a request. This Provisioned Concurrency initialization behavior may affect how your function interacts with data you store in ephemeral storage, even when your function isn’t processing requests. To learn more about Provisioned Concurrency, please see the relevant documentation.

You can configure each Lambda function with its own ephemeral storage between 512MB and 10,240MB, in 1MB increments by using the AWS Lambda console, AWS Lambda API, or AWS CloudFormation template during function creation or update.

Yes. All data stored in ephemeral storage is encrypted at rest with a key managed by AWS.

You can use AWS CloudWatch Lambda Insight metrics to monitor your ephemeral storage usage. To learn more, see the AWS CloudWatch Lambda Insights documentation.

Keeping functions stateless enables AWS Lambda to rapidly launch as many copies of the function as needed to scale to the rate of incoming events. While AWS Lambda’s programming model is stateless, your code can access stateful data by calling other web services, such as Amazon S3 or Amazon DynamoDB.

Yes. AWS Lambda allows you to use normal language and operating system features, such as creating additional threads and processes. Resources allocated to the Lambda function, including memory, execution time, disk, and network use, must be shared among all the threads/processes it uses. You can launch processes using any language supported by Amazon Linux.

Lambda attempts to impose as few restrictions as possible on normal language and operating system activities, but there are a few activities that are disabled: Inbound network connections are blocked by AWS Lambda, and for outbound connections, only TCP/IP and UDP/IP sockets are supported, and ptrace (debugging) system calls are blocked. TCP port 25 traffic is also blocked as an anti-spam measure.

If you are using Node.js or Python, you can author the code for your function using code editor in the AWS Lambda console, which lets you author and test your functions, and view the results of function executions in a robust, IDE-like environment. Go to the console to get started.

You can also package the code (and any dependent libraries) as a ZIP and upload it using the AWS Lambda console from your local environment or specify an Amazon S3 location where the ZIP file is located. Uploads must be no larger than 50MB (compressed). You can use the AWS Eclipse plugin to author and deploy Lambda functions in Java. You can use the Visual Studio plugin to author and deploy Lambda functions in C#, and Node.js.

You can package the code (and any dependent libraries) as a ZIP and upload it using the AWS CLI from your local environment, or specify an Amazon S3 location where the ZIP file is located. Uploads must be no larger than 50MB (compressed). Visit the Lambda Getting Started guide to get started.

Yes. You can easily create and modify environment variables from the AWS Lambda Console, CLI, or SDKs. To learn more about environment variables, see the documentation.

For sensitive information, such as database passwords, we recommend you use client-side encryption using AWS Key Management Service and store the resulting values as ciphertext in your environment variable. You will need to include logic in your AWS Lambda function code to decrypt these values.

You can adjust and secure the resources associated with your Lambda function using the Lambda API or console. To learn more about it, see the documentation.

Yes, you can package any code (frameworks, SDKs, libraries, and more) as a Lambda Layer and manage and share them easily across multiple functions.

AWS Lambda automatically monitors Lambda functions on your behalf, reporting real-time metrics through Amazon CloudWatch, including total requests, account-level and function-level concurrency usage, latency, error rates, and throttled requests. You can view statistics for each of your Lambda functions via the Amazon CloudWatch console or through the AWS Lambda console. You can also call third-party monitoring APIs in your Lambda function.
 

Visit Troubleshooting CloudWatch metrics to learn more. Standard charges for AWS Lambda apply to use Lambda’s built-in metrics.

AWS Lambda automatically integrates with Amazon CloudWatch logs, creating a log group for each Lambda function and providing basic application lifecycle event log entries, including logging the resources consumed for each use of that function. You can easily insert additional logging statements into your code. You can also call third-party logging APIs in your Lambda function. Visit Troubleshooting Lambda functions to learn more. Amazon CloudWatch Logs rates will apply.

You do not have to scale your Lambda functions – AWS Lambda scales them automatically on your behalf. Every time an event notification is received for your function, AWS Lambda quickly locates free capacity within its compute fleet and runs your code. Since your code is stateless, AWS Lambda can start as many copies of your function as needed without lengthy deployment and configuration delays. There are no fundamental limits to scaling a function. AWS Lambda will dynamically allocate capacity to match the rate of incoming events.

In the AWS Lambda resource model, you choose the amount of memory you want for your function, and are allocated proportional CPU power and other resources. For example, choosing 256MB of memory allocates approximately twice as much CPU power to your Lambda function as requesting 128MB of memory and half as much CPU power as choosing 512MB of memory. To learn more, see our Function Configuration documentation.

You can set your memory from 128MB to 10,240MB.

Customers running memory or compute-intensive workloads can now use more memory for their functions. Larger memory functions help multithreaded applications run faster, making them ideal for data and computationally intensive applications like machine learning, batch and ETL jobs, financial modeling, genomics, HPC, and media processing.

AWS Lambda functions can be configured to run up to 15 minutes per execution. You can set the timeout to any value between 1 second and 15 minutes.

AWS Lambda is priced on a pay-per-use basis. Please see the AWS Lambda pricing page for details.

Yes. In addition to saving money on Amazon EC2 and AWS Fargate, you can also use Compute Savings Plans to save money on AWS Lambda. Compute Savings Plans offer up to 17% discount on Duration, Provisioned Concurrency, and Duration (Provisioned Concurrency). Compute Savings Plans do not offer a discount on Requests in your Lambda bill. However, your Compute Savings Plans commitment can apply to Requests at regular rates.

Yes. By default, each AWS Lambda function has a single, current version of the code. Clients of your Lambda function can call a specific version or get the latest implementation. Please read our documentation on versioning Lambda functions.

Deployment times may vary with the size of your code, but AWS Lambda functions are typically ready to call within seconds of upload.

Yes. You can include your own copy of a library (including the AWS SDK) in order to use a different version than the default one provided by AWS Lambda.

AWS Lambda offers discounted pricing tiers for monthly on-demand function duration above certain thresholds. Tiered pricing is available for functions running on both x86 and Arm architectures. Lambda pricing tiers are applied to aggregate monthly on-demand duration of your functions running on the same architecture (x86 or Arm, respectively), in the same region, within the account. If you’re using consolidated billing in AWS Organizations, pricing tiers are applied to the aggregate monthly duration of your functions running on the same architecture, in the same region, across the accounts in the organization. For example, if you are running x86 Lambda functions in the US East (Ohio) region, you will pay $0.0000166667 for every GB-second for the first 6 billion GB-seconds per month, $0.0000150000 for every GB-second for the next 9 billion GB-seconds per month, and $0.0000133334 for every GB-second over 15 billion GB-seconds per month, in that region. Pricing for Requests, Provisioned Concurrency, and Provisioned Concurrency Duration remains unchanged. For more information, please see AWS Lambda Pricing

Yes. Lambda usage that is covered by your hourly savings plan commitment is billed at the applicable CSP rate and discount. The remaining usage that is not covered by this commitment will be billed at the rate corresponding to the tier your monthly aggregate function duration falls in.

An event source is an AWS service or developer-created application that produces events that trigger an AWS Lambda function to run. Some services publish these events to Lambda by invoking the cloud function directly (for example, Amazon S3). Lambda can also poll resources in other services that do not publish events to Lambda. For example, Lambda can pull records from an Amazon Kinesis stream or an Amazon SQS queue and execute a Lambda function for each fetched message.Many other services, such as AWS CloudTrail, can act as event sources simply by logging to Amazon S3 and using S3 bucket notifications to trigger AWS Lambda functions

Please see our documentation for a complete list of event sources.

Events are passed to a Lambda function as an event input parameter. For event sources where events arrive in batches, such as Amazon SQS, Amazon Kinesis, and Amazon DynamoDB Streams, the event parameter may contain multiple events in a single call, based on the batch size you request. To learn more about Amazon S3 event notifications, visit Configuring Notifications for Amazon S3 Events. To learn more about Amazon DynamoDB Streams, visit the DynamoDB Stream Developers Guide. To learn more about invoking Lambda functions using Amazon SNS, visit the Amazon SNS Developers Guide. For more information on Amazon Cognito events, visit Amazon Cognito. For more information on AWS CloudTrail logs and auditing API calls across AWS services, see AWS CloudTrail.

From the AWS Lambda console, you can select a function and associate it with notifications from an Amazon S3 bucket. Alternatively, you can use the Amazon S3 console and configure the bucket’s notifications to send to your AWS Lambda function. This same functionality is also available through the AWS SDK and CLI.

You can trigger a Lambda function on DynamoDB table updates by subscribing your Lambda function to the DynamoDB Stream associated with the table. You can associate a DynamoDB Stream with a Lambda function using the Amazon DynamoDB console, the AWS Lambda console, or Lambda’s registerEventSource API.

From the AWS Lambda console, you can select a Lambda function and associate it with an Amazon Kinesis stream owned by the same account. This same functionality is also available through the AWS SDK and CLI.

The Amazon Kinesis and DynamoDB Streams records sent to your AWS Lambda function are strictly serialized, per shard. This means that if you put two records in the same shard, Lambda guarantees that your Lambda function will be successfully invoked with the first record before it is invoked with the second record. If the invocation for one record times out, is throttled, or encounters any other error, Lambda will retry until it succeeds (or the record reaches its 24-hour expiration) before moving on to the next record. The ordering of records across different shards is not guaranteed, and processing of each shard happens in parallel.

AWS Lambda allows you to perform time-based aggregations (such as count, max, sum, average, etc.) over a short window of up to 15 minutes for your data in Amazon Kinesis or Amazon DynamoDB Streams over a single logical partition such as a shard. This gives you the option to easily set up simple analytics for your event-based application without adding architectural complexity, as your business and analytics logic can be located in the same function. Lambda allows aggregations over a maximum of a 15-minute tumbling window, based on the event timestamp. Amazon Kinesis Data Analytics allows you to build more complex analytics applications that support flexible processing choices and robust fault-tolerance with exactly-once processing without duplicates, and analytics that can be performed over an entire data stream across multiple logical partitions. With KDA, you can analyze data over multiple types of aggregation windows (tumbling window, stagger window, sliding window, session window) using either the event time or the processing time.
 

	AWS Lambda	Amazon KDA
Tumbling Window	Yes	Yes
Stagger Window	No	Yes
Sliding Window	No	Yes
Session Window	No	Yes
Enrichment	No	Yes
Joint input and reference tables	No	Yes
Split input stream	No	Yes
Exactly-once processing	No	Yes
Maximum time window	15 mins	No limit
Aggregation scope	Partition/shard	Stream
Time semantics	Event time	Event time, Processing time

From the AWS Lambda console, you can select a Lambda function and associate it with an Amazon SNS topic. This same functionality is also available through the AWS SDK and CLI.

From the Amazon SES Console, you can set up your receipt rule to have Amazon SES deliver your messages to an AWS Lambda function. The same functionality is available through the AWS SDK and CLI.

First, configure the alarm to send Amazon SNS notifications. Then from the AWS Lambda console, select a Lambda function and associate it with that Amazon SNS topic. See the Amazon CloudWatch Developer Guide for more on setting up Amazon CloudWatch alarms.

From the AWS Lambda console, you can select a function to trigger when any datasets associated with an Amazon Cognito identity pool are synchronized. This same functionality is also available through the AWS SDK and CLI. Visit Amazon Cognito for more information on using Amazon Cognito to share and synchronize data across a user’s devices.

You can invoke a Lambda function using a custom event through AWS Lambda’s invoke API. Only the function owner or another AWS account that the owner has granted permission can invoke the function. Visit the Lambda Developers Guide to learn more.

AWS Lambda is designed to process events within milliseconds. Latency will be higher immediately after a Lambda function is created, updated, or if it has not been used recently.

You upload the code you want AWS Lambda to execute and then invoke it from your mobile app using the AWS Lambda SDK included in the AWS Mobile SDK. You can make both direct (synchronous) calls to retrieve or check data in real time, as well as asynchronous calls. You can also define a custom API using Amazon API Gateway and invoke your Lambda functions through any REST compatible client. To learn more about the AWS Mobile SDK, visit the AWS Mobile SDK page. To learn more about Amazon API Gateway, visit the Amazon API Gateway page.

You can invoke a Lambda function over HTTPS by defining a custom RESTful API using Amazon API Gateway. This gives you an endpoint for your function which can respond to REST calls like GET, PUT, and POST. Read more about using AWS Lambda with Amazon API Gateway.

When called through the AWS Mobile SDK, AWS Lambda functions automatically gain insight into the device and application that made the call through the ‘context’ object.

When your app uses the Amazon Cognito identity, end users can authenticate themselves using a variety of public login providers such as Amazon, Facebook, Google, and other OpenID Connect-compatible services. User identity is then automatically and secured presented to your Lambda function in the form of an Amazon Cognito id, allowing it to access user data from Amazon Cognito, or as a key to store and retrieve data in Amazon DynamoDB or other web services.

AWS Lambda is integrated with the Alexa Skills Kit, a collection of self-service APIs, tools, documentation, and code samples that make it easy for you to create voice-driven capabilities (or “skills”) for Alexa. You simply upload the Lambda function code for the new Alexa skill you are creating, and AWS Lambda does the rest, executing the code in response to Alexa voice interactions and automatically managing the compute resources on your behalf. Read the Alexa Skills Kit documentation for more details.

For Amazon S3 bucket notifications and custom events, AWS Lambda will attempt execution of your function three times in the event of an error condition in your code or if you exceed a service or resource limit. For ordered event sources that AWS Lambda polls on your behalf, such as Amazon DynamoDB Streams and Amazon Kinesis streams, Lambda will continue attempting execution in the event of a developer code error until the data expires. You can monitor progress through the Amazon Kinesis and Amazon DynamoDB consoles and through the Amazon CloudWatch metrics that AWS Lambda generates for your function. You can also set Amazon CloudWatch alarms based on error or execution throttling rates.

Lambda-based applications (also referred to as serverless applications) are composed of functions triggered by events. A typical serverless application consists of one or more functions triggered by events such as object uploads to Amazon S3, Amazon SNS notifications, or API actions. These functions can stand alone or leverage other resources such as DynamoDB tables or Amazon S3 buckets. The most basic serverless application is simply a function.

You can deploy and manage your serverless applications using the AWS Serverless Application Model (AWS SAM). AWS SAM is a specification that prescribes the rules for expressing serverless applications on AWS. This specification aligns with the syntax used by AWS CloudFormation today and is supported natively within AWS CloudFormation as a set of resource types (referred to as "serverless resources"). These resources make it easier for AWS customers to use CloudFormation to configure and deploy serverless applications using existing CloudFormation APIs.

You can choose from a collection of serverless applications published by developers, companies, and partners in the AWS community with the AWS Serverless Application Repository. After finding an application, you can configure and deploy it straight from the Lambda console.

You can automate your serverless application release process using AWS CodePipeline and AWS CodeDeploy. CodePipeline is a continuous delivery service that enables you to model, visualize and automate the steps required to release your serverless application. CodeDeploy provides a deployment automation engine for your Lambda-based applications. CodeDeploy lets you orchestrate deployments according to established best-practice methodologies such as canary and linear deployments, and helps you establish the necessary guardrails to verify that newly-deployed code is safe, stable, and ready to be fully released to production.
 

To learn more about serverless CI/CD, visit our documentation.

To get started, visit the AWS Lambda console and download one of our blueprints. The file you download will contain an AWS SAM file (which defines the AWS resources in your application) and a .ZIP file (which includes your function code). You can then use AWS CloudFormation commands to package and deploy the serverless application that you just downloaded. For more details, visit our documentation.

You can use AWS Step Functions to coordinate a series of AWS Lambda functions in a specific order. You can invoke multiple Lambda functions sequentially, passing the output of one to the other, and/or in parallel, and Step Functions will maintain state during executions for you.

You can enable your Lambda function for tracing with AWS X-Ray by adding X-Ray permissions to your Lambda function execution role and changing your function “tracing mode” to “active. ” When X-Ray is enabled for your Lambda function, AWS Lambda will emit tracing information to X-Ray regarding the Lambda service overhead incurred when invoking your function. This will provide you with insights such as Lambda service overhead, function init time, and function execution time. In addition, you can include the X-Ray SDK in your Lambda deployment package to create your own trace segments, annotate your traces, or view trace segments for downstream calls made from your Lambda function. X-Ray SDKs are currently available for Node.js and Java. Visit Troubleshooting Lambda-based applications to learn more. AWS X-Ray rates will apply.

Yes. You can build highly scalable, secure, Lambda-based serverless applications that connect to relational databases using Amazon RDS Proxy, a highly available database proxy that manages thousands of concurrent connections to relational databases. Currently, RDS Proxy supports MySQL and Aurora databases. You can begin using RDS Proxy through the Amazon RDS console or the AWS Lambda console. Serverless applications that use fully managed connection pools from RDS Proxy will be billed according to RDS Proxy Pricing.

The specification is open sourced under Apache 2.0, which allows you and others to adopt and incorporate AWS SAM into build, deployment, monitoring, and management tools with a commercial-friendly license. You can access the AWS SAM repository on GitHub here.

AWS Lambda now enables you to package and deploy functions as container images. Customers can leverage the flexibility and familiarity of container tooling, and the agility and operational simplicity of AWS Lambda to build applications.

You can start with either an AWS provided base images for Lambda or by using one of your preferred community or private enterprise images. Then, simply use Docker CLI to build the image, upload it to Amazon ECR, and then create the function by using all familiar Lambda interfaces and tools, such as the AWS Management Console, the AWS CLI, the AWS SDK, AWS SAM, and AWS CloudFormation.

You can deploy third-party Linux base images (e.g. Alpine or Debian) to Lambda in addition to the Lambda provided images. AWS Lambda will support all images based on the following image manifest formats: Docker Image Manifest V2 Schema 2 (used with Docker version 1.10 and newer) or Open Container Initiative (OCI) Spec (v1.0 and up). Lambda supports images with a size of up to 10GB.

AWS Lambda provides a variety of base images customers can extend, and customers can also use their preferred Linux-based images with a size of up to 10GB.

You can use any container tooling as long as it supports one of the following container image manifest formats: Docker Image Manifest V2 Schema 2 (used with Docker version 1.10 and newer) or Open Container Initiative (OCI) Specifications (v1.0 and up). For example, you can use native container tools (i.e. docker run, docker compose, Buildah and Packer) to define your functions as a container image and deploy to Lambda.

All existing AWS Lambda features, with the exception of Lambda layers and Code Signing, can be used with functions deployed as container images. Once deployed, AWS Lambda will treat an image as immutable. Customers can use container layers during their build process to include dependencies.

Not at this time. Your image, once deployed to AWS Lambda, will be immutable. The service will not patch or update the image. However, AWS Lambda will publish curated base images for all supported runtimes that are based on the Lambda managed environment. These published images will be patched and updated along with updates to the AWS Lambda managed runtimes. You can pull and use the latest base image from DockerHub or Amazon ECR Public, re-build your container image and deploy to AWS Lambda via Amazon ECR. This allows you to build and test the updated images and runtimes, prior to deploying the image to production

There are three main differences between functions created using ZIP archives vs. container images:

1. Functions created using ZIP archives have a maximum code package size of 250 MB unzipped, and those created using container images have a maximum image size of 10 GB. 
2. Lambda uses Amazon ECR as the underlying code storage for functions defined as container images, so a function may not be invocable when the underlying image is deleted from ECR. 
3. ZIP functions are automatically patched for the latest runtime security and bug fixes. Functions defined as container images are immutable, and customers are responsible for the components packaged in their function. Customers can leverage the AWS provided base images which are regularly updated by AWS for security and bug fixes, using the most recent patches available.

No - AWS Lambda ensures that the performance profiles for functions packaged as container images are the same as for those packaged as ZIP archives, including typically sub-second start up times.

There is no additional charge for packaging and deploying functions as container images to AWS Lambda. When you invoke your function deployed as a container image, you pay the regular price for requests and execution duration. To learn more, visit AWS Lambda pricing. You will be charged for storing your container images in Amazon ECR at the standard ECR prices. To learn more, visit Amazon ECR pricing.

The Lambda Runtime Interface Emulator is a proxy for the Lambda Runtime API,which allows customers to locally test their Lambda function packaged as a container image. It is a lightweight web server that converts HTTP requests to JSON events and emulates the Lambda Runtime API. It allows you to locally test your functions using familiar tools such as cURL and the Docker CLI (when testing functions packaged as container images). It also simplifies running your application on additional compute services. You can include the Lambda Runtime Interface Emulator in your container image to have it accept HTTP requests natively instead of the JSON events required for deployment to Lambda. This component does not emulate the Lambda orchestrator, or security and authentication configurations. The Runtime Interface Emulator is open sourced on GitHub. You can get started by downloading and installing it on your local machine.

The Lambda Runtime API in the running Lambda service accepts JSON events and returns responses. The Lambda Runtime Interface Emulator allows the function packaged as a container image to accept HTTP requests during local testing with tools like cURL, and surface them via the same interface locally to the function. It allows you to use the docker run or docker-compose up command to locally test your lambda application.

You can use the emulator to test if your function code is compatible with the Lambda environment, runs successfully, and provides the expected output. For example, you can mock test events from different event sources. You can also use it to test extensions and agents built into the container image against the Lambda Extensions API.

Customers can add the Runtime Interface Emulator as the entry point to the container image or package it as a sidecar to ensure the container image now accepts HTTP requests instead of JSON events. This simplifies the changes required to run their container image on additional compute services. Customers will be responsible for ensuring they follow all security, performance, and concurrency best practices for their chosen environment. RIE is pre-packaged into the AWS Lambda provided images, and is available by default in AWS SAM CLI. Base image providers can use the documentation to provide the same experience for their base images.

You can deploy a containerized application to AWS Lambda if it meets the below requirements:

1. The container image must implement the Lambda Runtime API. We have open-sourced a set of software packages, Runtime Interface Clients (RIC), that implement the Lambda Runtime API, allowing you to seamlessly extend your preferred base images to be Lambda compatible.
2. The container image must be able to run on a read-only filesystem. Your function code can access a writable /tmp directory storage of 512 MB. If you are using an image that requires a writable root directory, configure it to write to the /tmp directory.
3. The files required for the execution of function code can be read by the default Lambda user. Lambda defines a default Linux user with least-privileged permissions that follows security best practices. You need to verify that your application code does not rely on files that are restricted by other Linux users for execution.
4. It is a Linux based container image.

AWS SnapStart can improve startup performance from several seconds to as low as sub-second for latency sensitive applications. SnapStart works by snapshotting your function's initialized memory (and disk) state and caching this snapshot for low-latency access. When your function is subsequently invoked, Lambda resumes execution environments from this pre-initialized snapshot instead of initializing them from scratch, improving startup latency. For resiliency, Lambda maintains cached copies of your snapshot and automatically applies software updates, such as runtime upgrades and security patches to them.

Lambda SnapStart is a simple function level configuration that can be configured for new and existing functions by using Lambda API, the AWS Management Console, AWS Command Line Interface (CLI), AWS SDK, AWS Cloud Development Kit (CDK), AWS CloudFormation, and the AWS Serverless Application Model (SAM). When you configure Lambda SnapStart, every function version that is published thereafter benefits from the improved startup performance offered by Lambda SnapStart. To learn more about Lambda SnapStart, see the documentation.

Lambda SnapStart is a performance optimization that helps your functions to achieve faster start-up times by reducing the variable latency incurred during execution of one-time initialization code. While Lambda SnapStart reduces startup latency, it works as a best-effort optimization, and does not guarantee elimination of cold starts. If your application has strict latency requirements and requires double-digit millisecond startup times, we recommend you use PC.

Lambda SnapStart supports multiple runtimes, including Java 11 (and newer), Python 3.12 (and newer) and .NET 8 (and newer). Future versions of runtimes will be supported after they are released. For all runtimes supported by Lambda, see the Lambda runtimes documentation.

No. Lambda SnapStart and PC cannot be enabled at the same time, on the same function.

Yes. You can configure a Lambda SnapStart function to access resources in a virtual private cloud (VPC). For more information on how to configure your function with a VPC, see the Lambda documentation.

Yes. You can configure Lambda SnapStart for functions running on both x86 and Arm architectures.

No. You cannot enable Lambda SnapStart with Amazon EFS at this time.

No. You cannot enable Lambda SnapStart with larger ephemeral storage (/tmp) beyond 512 MB at this time.

Yes. If your code assumes uniqueness of state, you need to evaluate your code’s resilience to snapshot operations (such as being cloned and resumed). To learn more on uniqueness considerations with Lambda SnapStart, see the documentation and blog on understanding uniqueness in VM snapshots with Lambda SnapStart.

Yes. You can implement your own software logic before creating (checkpointing) a snapshot and after restoring a snapshot using runtime hooks. To learn more, see the Lambda SnapStart documentation.

Yes, you will be charged for caching a snapshot over the period that your function version is active, for a minimum of 3 hours and per milli-second thereafter. The price depends on the amount of memory you allocate to your function. You are also charged each time Lambda resumes an execution environment by restoring your snapshot, with the price depending on the amount of memory you allocate to your function. To learn more about pricing for SnapStart, visit AWS Lambda Pricing.

SnapStart pricing does not apply to supported Java managed runtimes, which can only cache a snapshot for up to 14 days.

Like all Lambda functions, duration charges apply to SnapStart functions. For functions using SnapStart, duration includes the time it takes for the runtime to load, any code that runs in a runtime hook, and the initialization code executed when creating snapshot copies for resilience.

With Lambda SnapStart for Python and .NET, your function snapshots remain active as long as your function is active. For Java functions, the snapshot associated with a published function expires if it remains inactive for more than 14 days.

Snapshots are encrypted be default with customer-unique AWS Key Management Service (KMS) keys owned and managed by the Lambda service. Customers can also encrypt snapshots using a KMS key owned and managed by the customer.

The maximum allowed initialization duration for Lambda SnapStart will match the execution timeout duration you have configured for your function. The maximum configurable execution timeout limit for a function is 15 minutes.

Provisioned Concurrency gives you greater control over the performance of your serverless applications. When enabled, Provisioned Concurrency keeps functions initialized and hyper-ready to respond in double-digit milliseconds.

You can configure concurrency on your function through the AWS Management Console, the Lambda API, the AWS CLI, and AWS CloudFormation. The simplest way to benefit from Provisioned Concurrency is by using AWS Auto Scaling. You can use Application Auto Scaling to configure schedules, or have Auto Scaling automatically adjust the level of Provisioned Concurrency in real time as demand changes. To learn more about Provisioned Concurrency, see the documentation.

You don’t need to make any changes to your code to use Provisioned Concurrency. It works seamlessly with all existing functions and runtimes. There is no change to the invocation and execution model of Lambda when using Provisioned Concurrency.

Provisioned Concurrency adds a pricing dimension, of ‘Provisioned Concurrency’, for keeping functions initialized. When enabled, you pay for the amount of concurrency that you configure and for the period of time that you configure it. When your function executes while Provisioned Concurrency is configured on it, you also pay for Requests and execution Duration. To learn more about the pricing of Provisioned Concurrency, see AWS Lambda Pricing.

Provisioned Concurrency is ideal for building latency-sensitive applications, such as web or mobile backends, synchronously invoked APIs, and interactive microservices. You can easily configure the appropriate amount of concurrency based on your application's unique demand. You can increase the amount of concurrency during times of high demand and lower it, or turn it off completely, when demand decreases.

If the concurrency of a function reaches the configured level, subsequent invocations of the function have the latency and scale characteristics of regular Lambda functions. You can restrict your function to only scale up to the configured level. Doing so prevents the function from exceeding the configured level of Provisioned Concurrency. This is a mechanism to prevent undesired variability in your application when demand exceeds the anticipated amount.

AWS Lambda allows you to run your functions on either x86-based or Arm-based processors. AWS Graviton2 processors are custom built by Amazon Web Services using 64-bit Arm Neoverse cores to deliver increased price performance for your cloud workloads. Customers get the same advantages of AWS Lambda, running code without provisioning or managing servers, automatic scaling, high availability, and only paying for the resources you consume.

AWS Lambda functions powered by Graviton2, using an Arm-based processor architecture designed by AWS, are designed to deliver up to 34% better price performance compared to functions running on x86 processors, for a variety of serverless workloads, such as web and mobile backends, data, and stream processing. With lower latency, up to 19% better performance, a 20% lower cost, and the highest power-efficiency currently available at AWS, Graviton2 functions can power mission critical serverless applications. Customers can configure both existing and new functions to target the Graviton2 processor. They can deploy functions running on Graviton2 as either zip files or container images.

You can configure functions to run on Graviton2 through the AWS Management Console, the AWS Lambda API, the AWS CLI, and AWS CloudFormation by setting the architecture flag to ‘arm64’ for your function.

There is no change between x86-based and Arm-based functions. Simply upload your code via the AWS Management Console, zip file, or container image, and AWS Lambda automatically runs your code when triggered, without requiring you to provision or manage infrastructure.

Interpreted languages like Python, Java, and Node generally do not require recompilation unless your code references libraries that use architecture specific components. In those cases, you would need to provide the libraries targeted to arm64. For more details, please see the Getting started with AWS Graviton page. Non-interpreted languages will require compiling your code to target arm64. While more modern compilers will produce compiled code for arm64, you will need to deploy it into an arm-based environment to test. To learn more about using Lambda functions with Graviton2, please see the documentation.

An application can contain functions running on both architectures. AWS Lambda allows you to change the architecture (‘x86_64’ or ‘arm64’) of your function’s current version. Once you create a specific version of your function, the architecture cannot be changed.

No. Each function version can only use a single container image.

Yes. Layers and extensions can be targeted to ‘x86_64’ or ‘arm64’ compatible architectures. The default architecture for functions and layers is ‘x86_64’.

At launch, customers can use Python, Node.js, Java, Ruby, .Net Core, Custom Runtime (provided.al2), and OCI Base images. To learn more, please see the AWS Lambda Runtimes.

AWS Lambda functions powered by AWS Graviton2 processors are 20% cheaper compared to x86-based Lambda functions. The Lambda free tier applies to AWS Lambda functions powered by x86 and Arm-based architectures.

Each workload is unique and we recommend customers test their functions to determine the price performance improvement they might see. To do that, we recommend using the AWS Lambda Power Tuning tool. We recommend starting with web and mobile backends, data, and stream processing when testing your workloads for potential price performance improvements.

With Amazon Elastic File System (Amazon EFS) for AWS Lambda, customers can securely read, write and persist large volumes of data at virtually any scale using a fully managed elastic NFS file system that can scale on demand without the need for provisioning or capacity management. Previously, developers added code to their functions to download data from S3 or databases to local temporary storage, limited to 512MB. With EFS for Lambda, developers don't need to write code to download data to temporary storage in order to process it.

Developers can easily connect an existing EFS file system to a Lambda function via an EFS Access Point by using the console, CLI, or SDK. When the function is first invoked, the file system is automatically mounted and made available to function code. You can learn more in the documentation.

Yes. Mount targets for Amazon EFS are associated with a subnet in a VPC. The AWS Lambda function needs to be configured to access that VPC.

Using EFS for Lambda is ideal for building machine learning applications or loading large reference files or models, processing or backing up large amounts of data, hosting web content, or developing internal build systems. Customers can also use EFS for Lambda to keep state between invocations within a stateful microservice architecture, in a Step Functions workflow, or sharing files between serverless applications and instance or container-based applications.

Yes. Data encryption in transit uses industry-standard Transport Layer Security (TLS) 1.2 to encrypt data sent between AWS Lambda functions and the Amazon EFS file systems.

Customers can provision Amazon EFS to encrypt data at rest. Data encrypted at rest is transparently encrypted while being written, and transparently decrypted while being read, so you don’t have to modify your applications. Encryption keys are managed by the AWS Key Management Service (KMS), eliminating the need to build and maintain a secure key management infrastructure.

There is no additional charge for using Amazon EFS for AWS Lambda. Customers pay the standard price for AWS Lambda and for Amazon EFS. When using Lambda and EFS in the same availability zone, customers are not charged for data transfer. However, if they use VPC peering for Cross-Account access, they will incur data transfer charges. To learn more, please see Pricing.

No. Each Lambda function will be able to access one EFS file system.

Yes. Amazon EFS supports Lambda functions, ECS and Fargate containers, and EC2 instances. You can share the same file system and use IAM policy and Access Points to control what each function, container, or instance has access to.  

Yes. Lambda functions can be configured with a function URL, a built-in HTTPS endpoint that can be invoked using the browser, curl, and any HTTP client. Function URLs are an easy way to get started building HTTPS accessible functions.

You can configure a function URL for your function through the AWS Management Console, the AWS Lambda API, the AWS CLI, AWS CloudFormation, and the AWS Serverless Application Model. Function URLs can be enabled on the $LATEST unqualified version of your function, or on any function alias. To learn more about configuring a function URL, see the documentation.

Lambda function URLs are secured with IAM authorization by default. You can choose to disable IAM authorization to create a public endpoint or if you plan to implement custom authorization as part of the function’s business logic.

You can easily invoke your function from your web browser by navigating to the Lambda URL, from your client application’s code using an HTTP library, or from the command line using curl.

Yes. Lambda function URLs can be enabled on a function or function alias. If no alias is specified, the URL will point to $LATEST by default. Function URLs cannot target an individual function version.

Custom domain names are not currently supported with function URLs. You can use a custom domain with your function URL by creating an Amazon CloudFront distribution and a CNAME to map your custom domain to your CloudFront distribution name. Then, map your CloudFront distribution domain name to be routed to your function URL as an origin.

Yes, function URLs can be used to invoke a Lambda function in a VPC.

There is no additional charge for using function URLs. You pay the standard price for AWS Lambda. To learn more, please see AWS Lambda Pricing.

Lambda@Edge allows you to run code across AWS locations globally without provisioning or managing servers, responding to end-users at the lowest network latency. You just upload your Node.js or Python code to AWS Lambda and configure your function to be triggered in response to Amazon CloudFront requests (i.e., when a viewer request lands, when a request is forwarded to or received back from the origin, and right before responding back to the end-user). The code is then ready to execute across AWS locations globally when a request for content is received, and scales with the volume of CloudFront requests globally. Learn more in our documentation.

To use Lambda@Edge, you just upload your code to AWS Lambda and associate a function version to be triggered in response to Amazon CloudFront requests. Your code must satisfy the Lambda@Edge service limits. Lambda@Edge supports Node.js and Python for global invocation by CloudFront events at this time. Learn more in our documentation.

Lambda@Edge is optimized for latency-sensitive use cases where your end viewers are distributed globally. All the information you need to make a decision should be available at the CloudFront edge, within the function and the request. This means that use cases where you are looking to make decisions on how to serve content based on user characteristics (e.g., location, client device, etc.) can now be executed and served close to your users without having to be routed back to a centralized server.

You can associate existing Lambda functions with CloudFront events for global invocation if the function satisfies the Lambda@Edge service requirements and limits. Read more here on how to update your function properties.

Your functions will automatically trigger in response to the following Amazon CloudFront events:

• Viewer Request - This event occurs when an end-user or a device on the Internet makes an HTTP(S) request to CloudFront, and the request arrives at the edge location closest to that user.
• Viewer Response - This event occurs when the CloudFront server at the edge is ready to respond to the end user or the device that made the request.
• Origin Request - This event occurs when the CloudFront edge server does not already have the requested object in its cache, and the viewer request is ready to be sent to your backend origin web server (e.g. Amazon EC2, or Application Load Balancer, or Amazon S3).
• Origin Response - This event occurs when the CloudFront server at the edge receives a response from your backend origin web server.

The difference is that API Gateway and Lambda are regional services. Using Lambda@Edge and Amazon CloudFront allows you to execute logic across multiple AWS locations based on where your end viewers are located.

AWS Lambda is designed to use replication and redundancy to provide high availability for both the service itself and for the Lambda functions it operates. There are no maintenance windows or scheduled downtimes for either.

Yes. When you update a Lambda function, there will be a brief window of time, typically less than a minute, when requests could be served by either the old or the new version of your function.

No. AWS Lambda is designed to run many instances of your functions in parallel. However, AWS Lambda has a default safety throttle, for the number of concurrent executions per account per region (visit here for info on default safety throttle limits). You can also control the maximum concurrent executions for individual AWS Lambda functions, which you can use to reserve a subset of your account concurrency limit for critical functions, or cap traffic rates to downstream resources.

If you wish to submit a request to increase the concurrent execution limit, you can use Service Quotas to request a limit increase request.

On exceeding the maximum concurrent executions limit, AWS Lambda functions being invoked synchronously will return a throttling error (429 error code). Lambda functions being invoked asynchronously can absorb reasonable bursts of traffic for approximately 15-30 minutes, after which incoming events will be rejected as throttled. In case the Lambda function is being invoked in response to Amazon S3 events, events rejected by AWS Lambda may be retained and retried by S3 for 24 hours. Events from Amazon Kinesis streams and Amazon DynamoDB streams are retried until the Lambda function succeeds or the data expires. Amazon Kinesis and Amazon DynamoDB Streams retain data for 24 hours.

The default maximum concurrent execution limit is applied at the account level. However, you can also set limits on individually functions as well (visit here for info on Reserved Concurrency).

Each synchronously invoked Lambda function can scale at a rate of up to 1000 concurrent executions every 10 seconds. While Lambda's scaling rate is suitable for most use cases, it is particularly ideal for those with predictable or unpredictable bursts of traffic. For example, SLA-bound data processing would require predictable yet rapid scaling to meet processing demand. Similarly, serving breaking news articles, or flash sales could drive unpredictable levels of traffic in a short period of time. Lambda's scaling rate can facilitate such use cases without additional configurations or tooling. Additionally, the concurrency scaling limit is a function-level limit, which means each function in your account scales independently of other functions.

On failure, Lambda functions being invoked synchronously will respond with an exception. Lambda functions being invoked asynchronously are retried at least 3 times. Events from Amazon Kinesis streams and Amazon DynamoDB streams are retried until the Lambda function succeeds or the data expires. Kinesis and DynamoDB Streams retain data for a minimum of 24 hours.

On exceeding the retry policy for asynchronous invocations, you can configure a “dead letter queue” (DLQ) into which the event will be placed; in the absence of a configured DLQ the event may be rejected. On exceeding the retry policy for stream-based invocations, the data would have already expired and therefore rejected.

You can configure an Amazon SQS queue or an Amazon SNS topic as your dead letter queue.

You grant permissions to your Lambda function to access other resources using an IAM role. AWS Lambda assumes the role while executing your Lambda function, so you always retain full, secure control of exactly which AWS resources it can use. Visit Setting up AWS Lambda to learn more about roles.

When you configure an Amazon S3 bucket to send messages to an AWS Lambda function, a resource policy rule will be created that grants access. Visit the Lambda Developer Guide to learn more about resource policies and access controls for Lambda functions.

Access controls are managed through the Lambda function role. The role you assign to your Lambda function also determines which resource(s) AWS Lambda can poll on its behalf. Visit the Lambda Developer Guide to learn more.

Access controls can be managed by the Lambda function role or a resource policy setting on the queue itself. If both policies are present, the more restrictive of the two permissions will be applied.

You can enable Lambda functions to access resources in your VPC by specifying the subnet and security group as part of your function configuration. Lambda functions configured to access resources in a particular VPC will not have access to the internet as a default configuration. To grant internet to these functions, use internet gateways. By default, Lambda functions communicate with resources in a dual-stack VPC over IPv4. You can configure your functions to access resources in a dual-stack VPC over IPv6. For more details on Lambda functions configured with VPC, see Lambda Private Networking with VPC.

Code Signing for AWS Lambda offers trust and integrity controls that enable you to verify that only unaltered code from approved developers is deployed in your Lambda functions. You can use AWS Signer, a fully-managed code signing service, to digitally sign code artifacts and configure your Lambda functions to verify the signatures at deployment. Code Signing for AWS Lambda is currently only available for functions packaged as ZIP archives.

You can create digitally signed code artifacts using a Signing Profile through the AWS Signer console, the Signer API, SAM CLI or AWS CLI. To learn more, please see the documentation for AWS Signer.

You can enable code signing by creating a Code Signing Configuration through the AWS Management Console, the Lambda API, the AWS CLI, AWS CloudFormation, and AWS SAM. Code Signing Configuration helps you specify the approved signing profiles and configure whether to warn or reject deployments if signature checks fail. Code Signing Configurations can be attached to individual Lambda functions to enable the code signing feature. Such functions now start verifying signatures at deployment.

AWS Lambda can perform the following signature checks at deployment:

• Corrupt signature - This occurs if the code artifact has been altered since signing.
• Mismatched signature - This occurs if the code artifact is signed by a signing profile that is not approved.
• Expired signature - This occurs if the signature is past the configured expiry date.
• Revoked signature - This occurs if the signing profile owner revokes the signing jobs.

To learn more, please see the AWS Lambda documentation.

Yes, you can enable code signing for existing functions by attaching a code signing configuration to the function. You can do this using the AWS Lambda console, the Lambda API, the AWS CLI, AWS CloudFormation, and AWS SAM.

There is no additional cost when using Code Signing for AWS Lambda. You pay the standard price for AWS Lambda. To learn more, please see Pricing.

To provide you a simplified and enhanced logging experience by default, AWS Lambda offers advanced logging controls such as the ability to natively capture Lambda function logs in JSON structured format, control the log level filtering of Lambda function logs without making any code changes, and customize the Amazon CloudWatch log group Lambda sends logs to.

You can capture Lambda function logs in JSON structured format without having to use your own logging libraries. JSON structured logs make it easier to search, filter, and analyze large volumes of log entries. You can control the log level filtering of Lambda function logs without making any code changes, which enables you to choose the required logging granularity level for Lambda functions without sifting through large volumes of logs when debugging and troubleshooting errors. You can also set which Amazon CloudWatch log group Lambda sends logs to, making it easier to aggregate logs from multiple functions within an application in one place. You can then apply security, governance, and retention policies to logs at the application level rather than individually to every function.

You can specify advanced logging controls for your Lambda functions using AWS Lambda API, AWS Lambda console, AWS CLI, AWS Serverless Application Model (SAM), and AWS CloudFormation. To learn more, visit the launch blog post for advanced logging controls or the Lambda Developer Guide.

Yes, you can use your own logging libraries to generate Lambda logs in JSON structured format. To ensure your logging libraries work seamlessly with Lambda’s native JSON structured logging capability, Lambda will not double-encode any logs generated by your function that are already JSON encoded. You can also use Powertools for AWS Lambda library to capture Lambda logs in JSON structured format.

There is no additional charge for using advanced logging controls on Lambda. You will continue to be charged for ingestion and storage of your Lambda logs by Amazon CloudWatch Logs. See CloudWatch pricing page for log pricing details.

You can use standard tools like Maven or Gradle to compile your Lambda function. Your build process should mimic the same build process you would use to compile any Java code that depends on the AWS SDK. Run your Java compiler tool on your source files and include the AWS SDK 1.9 or later with transitive dependencies on your classpath. For more details, see our documentation.

Lambda provides the Amazon Linux build of openjdk 1.8.

Yes. You can use NPM packages as well as custom packages. Learn more here.

Yes. Lambda’s built-in sandbox lets you run batch (“shell”) scripts, other language runtimes, utility routines, and executables. Learn more here.

Yes. Any statically linked native module can be included in the ZIP file you upload, as well as dynamically linked modules compiled with an rpath pointing to your Lambda function root directory. Learn more here.

Yes. You can use Node.js' child_process command to execute a binary that you included in your function or any executable from Amazon Linux that is visible to your function. Alternatively several NPM packages exist that wrap command line binaries such as node-ffmpeg. Learn more here.

To deploy a Lambda function written in Node.js, simply package your Javascript code and dependent libraries as a ZIP. You can upload the ZIP from your local environment, or specify an Amazon S3 location where the ZIP file is located. For more details, see our documentation.

Yes. You can use pip to install any Python packages needed.

You can create a C# Lambda function using the Visual Studio IDE by selecting "Publish to AWS Lambda" in the Solution Explorer. Alternatively, you can directly run the "dotnet lambda publish" command from the dotnet CLI, which has the [# Lambda CLI tools patch] installed, which creates a ZIP of your C# source code along with all NuGet dependencies as well as your own published DLL assemblies, and automatically uploads it to AWS Lambda using the runtime parameter “dotnetcore1.0”

A PowerShell Lambda deployment package is a ZIP file that contains your PowerShell script, PowerShell modules that are required for your PowerShell script, and the assemblies needed to host PowerShell Core. You then use the AWSLambdaPSCore PowerShell module that you can install from the PowerShell Gallery to create your PowerShell Lambda deployment package.

Upload your Go executable artifact as a ZIP file through the AWS CLI or Lambda console and select the go1.x runtime. With Lambda, you can use Go's native tools to build and package your code. For more details, read our documentation. 

To deploy a Lambda function written in Ruby, package your Ruby code and gems as a ZIP. You can upload the ZIP from your local environment, or specify an Amazon S3 location where the ZIP file is located.

You can view the list of supported versions here.

No. AWS Lambda offers a single version of the operating system and managed language runtime to all users of the service. You can bring your own language runtime to use in Lambda.

AWS Lambda is integrated with AWS CloudTrail. AWS CloudTrail can record and deliver log files to your Amazon S3 bucket describing the API usage of your account.

You can use Amazon Step Functions to coordinate multiple invoking Lambda functions. You can invoke multiple Lambda functions serially, passing the output of one to the other, or in parallel. See our documentation for more details.

Yes, AWS Lambda supports the Advanced Vector Extensions 2 (AVX2) instruction set. To learn more about how to compile your application code to target this instruction set for improved performance, visit the AWS Lambda developer documentation.