Using AWS Lambda with Amazon Kinesis

You can use an AWS Lambda function to process records in an Amazon Kinesis data stream.

A Kinesis data stream is a set of shards. Each shard contains a sequence of data records. A consumer is an application that processes the data from a Kinesis data stream. You can map a Lambda function to a shared-throughput consumer (standard iterator), or to a dedicated-throughput consumer with enhanced fan-out.

For standard iterators, Lambda polls each shard in your Kinesis stream for records using HTTP protocol. The event source mapping shares read throughput with other consumers of the shard.

To minimize latency and maximize read throughput, you can create a data stream consumer with enhanced fan-out. Stream consumers get a dedicated connection to each shard that doesn't impact other applications reading from the stream. The dedicated throughput can help if you have many applications reading the same data, or if you're reprocessing a stream with large records. Kinesis pushes records to Lambda over HTTP/2.

For details about Kinesis data streams, see Reading Data from Amazon Kinesis Data Streams.

Example event

{
    "Records": [
        {
            "kinesis": {
                "kinesisSchemaVersion": "1.0",
                "partitionKey": "1",
                "sequenceNumber": "49590338271490256608559692538361571095921575989136588898",
                "data": "SGVsbG8sIHRoaXMgaXMgYSB0ZXN0Lg==",
                "approximateArrivalTimestamp": 1545084650.987
            },
            "eventSource": "aws:kinesis",
            "eventVersion": "1.0",
            "eventID": "shardId-000000000006:49590338271490256608559692538361571095921575989136588898",
            "eventName": "aws:kinesis:record",
            "invokeIdentityArn": "arn:aws:iam::123456789012:role/lambda-role",
            "awsRegion": "us-east-2",
            "eventSourceARN": "arn:aws:kinesis:us-east-2:123456789012:stream/lambda-stream"
        },
        {
            "kinesis": {
                "kinesisSchemaVersion": "1.0",
                "partitionKey": "1",
                "sequenceNumber": "49590338271490256608559692540925702759324208523137515618",
                "data": "VGhpcyBpcyBvbmx5IGEgdGVzdC4=",
                "approximateArrivalTimestamp": 1545084711.166
            },
            "eventSource": "aws:kinesis",
            "eventVersion": "1.0",
            "eventID": "shardId-000000000006:49590338271490256608559692540925702759324208523137515618",
            "eventName": "aws:kinesis:record",
            "invokeIdentityArn": "arn:aws:iam::123456789012:role/lambda-role",
            "awsRegion": "us-east-2",
            "eventSourceARN": "arn:aws:kinesis:us-east-2:123456789012:stream/lambda-stream"
        }
    ]
}

Polling and batching streams

Lambda reads records from the data stream and invokes your function synchronously with an event that contains stream records. Lambda reads records in batches and invokes your function to process records from the batch. Each batch contains records from a single shard/data stream.

By default, Lambda invokes your function as soon as records are available. If the batch that Lambda reads from the event source has only one record in it, Lambda sends only one record to the function. To avoid invoking the function with a small number of records, you can tell the event source to buffer records for up to 5 minutes by configuring a batching window. Before invoking the function, Lambda continues to read records from the event source until it has gathered a full batch, the batching window expires, or the batch reaches the payload limit of 6 MB. For more information, see Batching behavior.

If your function returns an error, Lambda retries the batch until processing succeeds or the data expires. To avoid stalled shards, you can configure the event source mapping to retry with a smaller batch size, limit the number of retries, or discard records that are too old. To retain discarded events, you can configure the event source mapping to send details about failed batches to a standard SQS queue or standard SNS topic.

To increase concurrency, you can also process multiple batches from each shard in parallel. Lambda can process up to 10 batches in each shard simultaneously. If you increase the number of concurrent batches per shard, Lambda still ensures in-order processing at the partition-key level.

Configure the ParallelizationFactor setting to process one shard of a Kinesis or DynamoDB data stream with more than one Lambda invocation simultaneously. You can specify the number of concurrent batches that Lambda polls from a shard via a parallelization factor from 1 (default) to 10. For example, when you set ParallelizationFactor to 2, you can have 200 concurrent Lambda invocations at maximum to process 100 Kinesis data shards (though in practice, you may see a different values for the ConcurrentExecutions metric). This helps scale up the processing throughput when the data volume is volatile and the IteratorAge is high.

You can also use the ParallelizationFactor with Kinesis aggregation. The behavior of the event source mapping depends on whether you're using enhanced fan-out:

Polling and stream starting position

Be aware that stream polling during event source mapping creation and updates is eventually consistent.

This behavior means that if you specify LATEST as the starting position for the stream, the event source mapping could miss events during creation or updates. To ensure that no events are missed, specify the stream starting position as TRIM_HORIZON or AT_TIMESTAMP.

Configuring your data stream and function

Your Lambda function is a consumer application for your data stream. It processes one batch of records at a time from each shard. You can map a Lambda function to a data stream (standard iterator), or to a consumer of a stream (enhanced fan-out).

For standard iterators, Lambda polls each shard in your Kinesis stream for records at a base rate of once per second. When more records are available, Lambda keeps processing batches until the function catches up with the stream. The event source mapping shares read throughput with other consumers of the shard.

To minimize latency and maximize read throughput, create a data stream consumer with enhanced fan-out. Enhanced fan-out consumers get a dedicated connection to each shard that doesn't impact other applications reading from the stream. Stream consumers use HTTP/2 to reduce latency by pushing records to Lambda over a long-lived connection and by compressing request headers. You can create a stream consumer with the Kinesis RegisterStreamConsumer API.

aws kinesis register-stream-consumer --consumer-name con1 \
--stream-arn arn:aws:kinesis:us-east-2:123456789012:stream/lambda-stream

You should see the following output:

{
    "Consumer": {
        "ConsumerName": "con1",
        "ConsumerARN": "arn:aws:kinesis:us-east-2:123456789012:stream/lambda-stream/consumer/con1:1540591608",
        "ConsumerStatus": "CREATING",
        "ConsumerCreationTimestamp": 1540591608.0
    }
}

To increase the speed at which your function processes records, add shards to your data stream. Lambda processes records in each shard in order. It stops processing additional records in a shard if your function returns an error. With more shards, there are more batches being processed at once, which lowers the impact of errors on concurrency.

If your function can't scale up to handle the total number of concurrent batches, request a quota increase or reserve concurrency for your function.

Execution role permissions

Lambda needs the following permissions to manage resources that are related to your Kinesis data stream. The AWSLambdaKinesisExecutionRole managed policy includes these permissions. You can add this managed policy to your function's execution role.

If your Kinesis data stream and Lambda function are in different accounts, ensure that your stream resource grants kinesis:DescribeStream permissions to your Lambda function's execution role or account.

Additionally, when creating your event source mapping from the console, you must have the kinesis:ListStreams and kinesis:ListStreamConsumers permissions.

To send records of failed batches to a standard SQS queue or standard SNS topic, your function needs additional permissions. Each destination service requires a different permission, as follows:

Add permissions and create the event source mapping

Create an event source mapping to tell Lambda to send records from your data stream to a Lambda function. You can create multiple event source mappings to process the same data with multiple Lambda functions, or to process items from multiple data streams with a single function. When processing items from multiple data streams, each batch will only contain records from a single shard/stream.

To configure your function to read from a Kinesis data stream, add the AWSLambdaKinesisExecutionRole AWS managed policy to your execution role and create a Kinesis trigger.

Lambda supports the following options for Kinesis event sources:

To manage the event source configuration later, choose the trigger in the designer.

Filtering Kinesis events

When you configure Kinesis as an event source for Lambda, you can use event filtering to control which records from your stream Lambda sends to your function for processing. To learn more about using Lambda event filtering with Kinesis, see Filtering with Kinesis.

Event source mapping API

To manage an event source with the AWS Command Line Interface (AWS CLI) or an AWS SDK, you can use the following API operations:

To create the event source mapping with the AWS CLI, use the create-event-source-mapping command. The following example uses the AWS CLI to map a function named my-function to a Kinesis data stream. The data stream is specified by an Amazon Resource Name (ARN), with a batch size of 500, starting from the timestamp in Unix time.

aws lambda create-event-source-mapping --function-name my-function \
--batch-size 500 --starting-position AT_TIMESTAMP --starting-position-timestamp 1541139109 \
--event-source-arn arn:aws:kinesis:us-east-2:123456789012:stream/lambda-stream

You should see the following output:

{
    "UUID": "2b733gdc-8ac3-cdf5-af3a-1827b3b11284",
    "BatchSize": 500,
    "MaximumBatchingWindowInSeconds": 0,
    "ParallelizationFactor": 1,
    "EventSourceArn": "arn:aws:kinesis:us-east-2:123456789012:stream/lambda-stream",
    "FunctionArn": "arn:aws:lambda:us-east-2:123456789012:function:my-function",
    "LastModified": 1541139209.351,
    "LastProcessingResult": "No records processed",
    "State": "Creating",
    "StateTransitionReason": "User action",
    "DestinationConfig": {},
    "MaximumRecordAgeInSeconds": 604800,
    "BisectBatchOnFunctionError": false,
    "MaximumRetryAttempts": 10000
}

To use a consumer, specify the consumer's ARN instead of the stream's ARN.

Configure additional options to customize how batches are processed and to specify when to discard records that can't be processed. The following example updates an event source mapping to send a failure record to a standard SQS queue after two retry attempts, or if the records are more than an hour old.

aws lambda update-event-source-mapping --uuid f89f8514-cdd9-4602-9e1f-01a5b77d449b \
--maximum-retry-attempts 2  --maximum-record-age-in-seconds 3600
--destination-config '{"OnFailure": {"Destination": "arn:aws:sqs:us-east-2:123456789012:dlq"}}'

You should see this output:

{
    "UUID": "f89f8514-cdd9-4602-9e1f-01a5b77d449b",
    "BatchSize": 100,
    "MaximumBatchingWindowInSeconds": 0,
    "ParallelizationFactor": 1,
    "EventSourceArn": "arn:aws:kinesis:us-east-2:123456789012:stream/lambda-stream",
    "FunctionArn": "arn:aws:lambda:us-east-2:123456789012:function:my-function",
    "LastModified": 1573243620.0,
    "LastProcessingResult": "PROBLEM: Function call failed",
    "State": "Updating",
    "StateTransitionReason": "User action",
    "DestinationConfig": {},
    "MaximumRecordAgeInSeconds": 604800,
    "BisectBatchOnFunctionError": false,
    "MaximumRetryAttempts": 10000
}

Updated settings are applied asynchronously and aren't reflected in the output until the process completes. Use the get-event-source-mapping command to view the current status.

aws lambda get-event-source-mapping --uuid f89f8514-cdd9-4602-9e1f-01a5b77d449b

You should see this output:

{
    "UUID": "f89f8514-cdd9-4602-9e1f-01a5b77d449b",
    "BatchSize": 100,
    "MaximumBatchingWindowInSeconds": 0,
    "ParallelizationFactor": 1,
    "EventSourceArn": "arn:aws:kinesis:us-east-2:123456789012:stream/lambda-stream",
    "FunctionArn": "arn:aws:lambda:us-east-2:123456789012:function:my-function",
    "LastModified": 1573244760.0,
    "LastProcessingResult": "PROBLEM: Function call failed",
    "State": "Enabled",
    "StateTransitionReason": "User action",
    "DestinationConfig": {
        "OnFailure": {
            "Destination": "arn:aws:sqs:us-east-2:123456789012:dlq"
        }
    },
    "MaximumRecordAgeInSeconds": 3600,
    "BisectBatchOnFunctionError": false,
    "MaximumRetryAttempts": 2
}

To process multiple batches concurrently, use the --parallelization-factor option.

aws lambda update-event-source-mapping --uuid 2b733gdc-8ac3-cdf5-af3a-1827b3b11284 \
--parallelization-factor 5

Error handling

The event source mapping that reads records from your Kinesis stream, invokes your function synchronously, and retries on errors. If Lambda throttles the function or returns an error without invoking the function, Lambda retries until the records expire or exceed the maximum age that you configure on the event source mapping.

If the function receives the records but returns an error, Lambda retries until the records in the batch expire, exceed the maximum age, or reach the configured retry quota. For function errors, you can also configure the event source mapping to split a failed batch into two batches. Retrying with smaller batches isolates bad records and works around timeout issues. Splitting a batch does not count towards the retry quota.

If the error handling measures fail, Lambda discards the records and continues processing batches from the stream. With the default settings, this means that a bad record can block processing on the affected shard for up to one week. To avoid this, configure your function's event source mapping with a reasonable number of retries and a maximum record age that fits your use case.

To retain a record of discarded batches, configure a failed-event destination. Lambda sends a document to the destination queue or topic with details about the batch.

The following example shows an invocation record for a Kinesis stream.

{
    "requestContext": {
        "requestId": "c9b8fa9f-5a7f-xmpl-af9c-0c604cde93a5",
        "functionArn": "arn:aws:lambda:us-east-2:123456789012:function:myfunction",
        "condition": "RetryAttemptsExhausted",
        "approximateInvokeCount": 1
    },
    "responseContext": {
        "statusCode": 200,
        "executedVersion": "$LATEST",
        "functionError": "Unhandled"
    },
    "version": "1.0",
    "timestamp": "2019-11-14T00:38:06.021Z",
    "KinesisBatchInfo": {
        "shardId": "shardId-000000000001",
        "startSequenceNumber": "49601189658422359378836298521827638475320189012309704722",
        "endSequenceNumber": "49601189658422359378836298522902373528957594348623495186",
        "approximateArrivalOfFirstRecord": "2019-11-14T00:38:04.835Z",
        "approximateArrivalOfLastRecord": "2019-11-14T00:38:05.580Z",
        "batchSize": 500,
        "streamArn": "arn:aws:kinesis:us-east-2:123456789012:stream/mystream"
    }
}

You can use this information to retrieve the affected records from the stream for troubleshooting. The actual records aren't included, so you must process this record and retrieve them from the stream before they expire and are lost.

Amazon CloudWatch metrics

Lambda emits the IteratorAge metric when your function finishes processing a batch of records. The metric indicates how old the last record in the batch was when processing finished. If your function is processing new events, you can use the iterator age to estimate the latency between when a record is added and when the function processes it.

An increasing trend in iterator age can indicate issues with your function. For more information, see Working with Lambda function metrics.

Time windows

Lambda functions can run continuous stream processing applications. A stream represents unbounded data that flows continuously through your application. To analyze information from this continuously updating input, you can bound the included records using a window defined in terms of time.

Tumbling windows are distinct time windows that open and close at regular intervals. By default, Lambda invocations are stateless—you cannot use them for processing data across multiple continuous invocations without an external database. However, with tumbling windows, you can maintain your state across invocations. This state contains the aggregate result of the messages previously processed for the current window. Your state can be a maximum of 1 MB per shard. If it exceeds that size, Lambda terminates the window early.

Each record in a stream belongs to a specific window. Lambda will process each record at least once, but doesn't guarantee that each record will be processed only once. In rare cases, such as error handling, some records might be processed more than once. Records are always processed in order the first time. If records are processed more than once, they might be processed out of order.

Aggregation and processing

Your user managed function is invoked both for aggregation and for processing the final results of that aggregation. Lambda aggregates all records received in the window. You can receive these records in multiple batches, each as a separate invocation. Each invocation receives a state. Thus, when using tumbling windows, your Lambda function response must contain a state property. If the response does not contain a state property, Lambda considers this a failed invocation. To satisfy this condition, your function can return a TimeWindowEventResponse object, which has the following JSON shape:

{
    "state": {
        "1": 282,
        "2": 715
    },
    "batchItemFailures": []
}

At the end of the window, the flag isFinalInvokeForWindow is set to true to indicate that this is the final state and that it’s ready for processing. After processing, the window completes and your final invocation completes, and then the state is dropped.

At the end of your window, Lambda uses final processing for actions on the aggregation results. Your final processing is synchronously invoked. After successful invocation, your function checkpoints the sequence number and stream processing continues. If invocation is unsuccessful, your Lambda function suspends further processing until a successful invocation.

{
    "Records": [
        {
            "kinesis": {
                "kinesisSchemaVersion": "1.0",
                "partitionKey": "1",
                "sequenceNumber": "49590338271490256608559692538361571095921575989136588898",
                "data": "SGVsbG8sIHRoaXMgaXMgYSB0ZXN0Lg==",
                "approximateArrivalTimestamp": 1607497475.000
            },
            "eventSource": "aws:kinesis",
            "eventVersion": "1.0",
            "eventID": "shardId-000000000006:49590338271490256608559692538361571095921575989136588898",
            "eventName": "aws:kinesis:record",
            "invokeIdentityArn": "arn:aws:iam::123456789012:role/lambda-kinesis-role",
            "awsRegion": "us-east-1",
            "eventSourceARN": "arn:aws:kinesis:us-east-1:123456789012:stream/lambda-stream"
        }
    ],
    "window": {
        "start": "2020-12-09T07:04:00Z",
        "end": "2020-12-09T07:06:00Z"
    },
    "state": {
        "1": 282,
        "2": 715
    },
    "shardId": "shardId-000000000006",
    "eventSourceARN": "arn:aws:kinesis:us-east-1:123456789012:stream/lambda-stream",
    "isFinalInvokeForWindow": false,
    "isWindowTerminatedEarly": false
}

Configuration

You can configure tumbling windows when you create or update an event source mapping. To configure a tumbling window, specify the window in seconds. The following example AWS Command Line Interface (AWS CLI) command creates a streaming event source mapping that has a tumbling window of 120 seconds. The Lambda function defined for aggregation and processing is named tumbling-window-example-function.

aws lambda create-event-source-mapping --event-source-arn arn:aws:kinesis:us-east-1:123456789012:stream/lambda-stream --function-name "arn:aws:lambda:us-east-1:123456789018:function:tumbling-window-example-function" --region us-east-1 --starting-position TRIM_HORIZON --tumbling-window-in-seconds 120

Lambda determines tumbling window boundaries based on the time when records were inserted into the stream. All records have an approximate timestamp available that Lambda uses in boundary determinations.

Tumbling window aggregations do not support resharding. When the shard ends, Lambda considers the window closed, and the child shards start their own window in a fresh state.

Tumbling windows fully support the existing retry policies maxRetryAttempts and maxRecordAge.

def lambda_handler(event, context):
    print('Incoming event: ', event)
    print('Incoming state: ', event['state'])

#Check if this is the end of the window to either aggregate or process.
    if event['isFinalInvokeForWindow']:
        # logic to handle final state of the window
        print('Destination invoke')
    else:
        print('Aggregate invoke')

#Check for early terminations
    if event['isWindowTerminatedEarly']:
        print('Window terminated early')

    #Aggregation logic
    state = event['state']
    for record in event['Records']:
        state[record['kinesis']['partitionKey']] = state.get(record['kinesis']['partitionKey'], 0) + 1

    print('Returning state: ', state)
    return {'state': state}

Reporting batch item failures

When consuming and processing streaming data from an event source, by default Lambda checkpoints to the highest sequence number of a batch only when the batch is a complete success. Lambda treats all other results as a complete failure and retries processing the batch up to the retry limit. To allow for partial successes while processing batches from a stream, turn on ReportBatchItemFailures. Allowing partial successes can help to reduce the number of retries on a record, though it doesn’t entirely prevent the possibility of retries in a successful record.

To turn on ReportBatchItemFailures, include the enum value ReportBatchItemFailures in the FunctionResponseTypes list. This list indicates which response types are enabled for your function. You can configure this list when you create or update an event source mapping.

Report syntax

When configuring reporting on batch item failures, the StreamsEventResponse class is returned with a list of batch item failures. You can use a StreamsEventResponse object to return the sequence number of the first failed record in the batch. You can also create your own custom class using the correct response syntax. The following JSON structure shows the required response syntax:

{ 
  "batchItemFailures": [ 
        {
            "itemIdentifier": "<SequenceNumber>"
        }
    ]
}

Success and failure conditions

Lambda treats a batch as a complete success if you return any of the following:

Lambda treats a batch as a complete failure if you return any of the following:

Lambda retries failures based on your retry strategy.

Bisecting a batch

If your invocation fails and BisectBatchOnFunctionError is turned on, the batch is bisected regardless of your ReportBatchItemFailures setting.

When a partial batch success response is received and both BisectBatchOnFunctionError and ReportBatchItemFailures are turned on, the batch is bisected at the returned sequence number and Lambda retries only the remaining records.

Here are some examples of function code that return the list of failed message IDs in the batch:

Amazon Kinesis configuration parameters

All Lambda event source types share the same CreateEventSourceMapping and UpdateEventSourceMapping API operations. However, only some of the parameters apply to Kinesis.