Configuring a Lambda function to stream responses

You can configure your Lambda function URLs to stream response payloads back to clients. Response streaming can benefit latency sensitive applications by improving time to first byte (TTFB) performance. This is because you can send partial responses back to the client as they become available. Additionally, you can use response streaming to build functions that return larger payloads. Response stream payloads have a soft limit of 20 MB as compared to the 6 MB limit for buffered responses. Streaming a response also means that your function doesn’t need to fit the entire response in memory. For very large responses, this can reduce the amount of memory you need to configure for your function.

The speed at which Lambda streams your responses depends on the response size. The streaming rate for the first 6MB of your function’s response is uncapped. For responses larger than 6MB, the remainder of the response is subject to a bandwidth cap. For more information on streaming bandwidth, see Bandwidth limits for response streaming.

Streaming responses incurs a cost. For more information, see Amazon Lambda Pricing.

Currently, Lambda supports response streaming only on Node.js 14.x, Node.js 16.x, and Node.js 18.x managed runtimes. You can also use a custom runtime with a custom Runtime API integration to stream responses or use the Lambda Web Adapter. You can stream responses through Lambda Function URLs, the Amazon SDK, or using the Lambda InvokeWithResponseStream API.

Writing response streaming-enabled functions

Writing the handler for response streaming functions is different than typical handler patterns. When writing streaming functions, be sure to do the following:

Configuring a handler function to stream responses

To indicate to the runtime that Lambda should stream your function's responses, you must wrap your function with the streamifyResponse() decorator. This tells the runtime to use the proper logic path for streaming responses and enables the function to stream responses.

The streamifyResponse() decorator accepts a function that accepts the following parameters:

The responseStream object is a Node.js writableStream. As with any such stream, you should use the pipeline() method.

const pipeline = require("util").promisify(require("stream").pipeline);
const { Readable } = require('stream');

exports.echo = awslambda.streamifyResponse(async (event, responseStream, _context) => {
  // As an example, convert event to a readable stream.
  const requestStream = Readable.from(Buffer.from(JSON.stringify(event)));

  await pipeline(requestStream, responseStream);
});

While responseStream offers the write() method to write to the stream, we recommend that you use pipeline() wherever possible. Using pipeline() ensures that the writable stream is not overwhelmed by a faster readable stream.

Ending the stream

Make sure that you properly end the stream before the handler returns. The pipeline() method handles this automatically.

For other use cases, call the responseStream.end() method to properly end a stream. This method signals that no more data should be written to the stream. This method isn't required if you write to the stream with pipeline() or pipe().

const pipeline = require("util").promisify(require("stream").pipeline);

exports.handler = awslambda.streamifyResponse(async (event, responseStream, _context) => {
  await pipeline(requestStream, responseStream);
});

exports.handler = awslambda.streamifyResponse(async (event, responseStream, _context) => {
  responseStream.write("Hello ");
  responseStream.write("world ");
  responseStream.write("from ");
  responseStream.write("Lambda!");
  responseStream.end();
});

Invoking a response streaming enabled function using Lambda function URLs

You can invoke response streaming enabled functions by changing the invoke mode of your function's URL. The invoke mode determines which API operation Lambda uses to invoke your function. The available invoke modes are:

You can still invoke your function without response streaming by directly calling the Invoke API operation. However, Lambda streams all response payloads for invocations that come through the function's URL until you change the invoke mode to BUFFERED.

To set the invoke mode of a function's URL (Amazon CLI)

aws lambda update-function-url-config --function-name my-function --invoke-mode RESPONSE_STREAM

To set the invoke mode of a function's URL (Amazon CloudFormation)

MyFunctionUrl:
  Type: AWS::Lambda::Url
  Properties:
    AuthType: AWS_IAM
    InvokeMode: RESPONSE_STREAM

For more information about configuring function URLs, see Lambda function URLs.

Bandwidth limits for response streaming

The first 6MB of your function’s response payload has uncapped bandwidth. After this initial burst, Lambda streams your response at a maximum rate of 2MBps. If your function responses never exceed 6MB, then this bandwidth limit never applies.

The rate of uncapped bandwidth varies depending on a number of factors, including your function’s processing speed. You can normally expect a rate higher than 2MBps for the first 6MB of your function’s response. If your function is streaming a response to a destination outside of Amazon, the streaming rate also depends on the speed of the external internet connection.