Example: Sliding Window - Managed Service for Apache Flink
Create Dependent ResourcesWrite Sample Records to the Input StreamDownload and Examine the Application CodeCompile the Application CodeUpload the Apache Flink Streaming Java CodeCreate and Run the Managed Service for Apache Flink ApplicationClean Up

Amazon Managed Service for Apache Flink was previously known as Amazon Kinesis Data Analytics for Apache Flink.

Example: Sliding Window

Note

To set up required prerequisites for this exercise, first complete the Getting Started (DataStream API) exercise.

Create Dependent Resources

Before you create a Managed Service for Apache Flink application for this exercise, you create the following dependent resources:

  • Two Kinesis data streams (ExampleInputStream and ExampleOutputStream).

  • An Amazon S3 bucket to store the application's code (ka-app-code-<username>)

You can create the Kinesis streams and Amazon S3 bucket using the console. For instructions for creating these resources, see the following topics:

  • Creating and Updating Data Streams in the Amazon Kinesis Data Streams Developer Guide. Name your data streams ExampleInputStream and ExampleOutputStream.

  • How Do I Create an S3 Bucket? in the Amazon Simple Storage Service User Guide. Give the Amazon S3 bucket a globally unique name by appending your login name, such as ka-app-code-<username>.

Write Sample Records to the Input Stream

In this section, you use a Python script to write sample records to the stream for the application to process.

Note

This section requires the AWS SDK for Python (Boto).

  1. Create a file named stock.py with the following contents:

     
import datetime
import json
import random
import boto3

STREAM_NAME = "ExampleInputStream"


def get_data():
    return {
        'EVENT_TIME': datetime.datetime.now().isoformat(),
        'TICKER': random.choice(['AAPL', 'AMZN', 'MSFT', 'INTC', 'TBV']),
        'PRICE': round(random.random() * 100, 2)}


def generate(stream_name, kinesis_client):
    while True:
        data = get_data()
        print(data)
        kinesis_client.put_record(
            StreamName=stream_name,
            Data=json.dumps(data),
            PartitionKey="partitionkey")


if __name__ == '__main__':
    generate(STREAM_NAME, boto3.client('kinesis'))

  2. Run the stock.py script: 

    $ python stock.py

    Keep the script running while completing the rest of the tutorial.

Download and Examine the Application Code

The Java application code for this example is available from GitHub. To download the application code, do the following:

  1. Install the Git client if you haven't already. For more information, see Installing Git.

  2. Clone the remote repository with the following command:

    git clone https://github.com/aws-samples/amazon-kinesis-data-analytics-examples.git

  3. Navigate to the amazon-kinesis-data-analytics-java-examples/SlidingWindow directory.

The application code is located in the SlidingWindowStreamingJobWithParallelism.java file. Note the following about the application code:

  • The application uses a Kinesis source to read from the source stream. The following snippet creates the Kinesis source:

    return env.addSource(new FlinkKinesisConsumer<>(inputStreamName,
                new SimpleStringSchema(), inputProperties));

  • The application uses the timeWindow operator to find the minimum value for each stock symbol over a 10-second window that slides by 5 seconds. The following code creates the operator and sends the aggregated data to a new Kinesis Data Streams sink:

  • Add the following import statement:

    import org.apache.flink.streaming.api.windowing.assigners.TumblingProcessingTimeWindows; //flink 1.13 onward

  • The application uses the timeWindow operator to find the count of values for each stock symbol over a 5-second tumbling window. The following code creates the operator and sends the aggregated data to a new Kinesis Data Streams sink:

    input.flatMap(new Tokenizer()) // Tokenizer for generating words
                .keyBy(0) // Logically partition the stream for each word
               
		.window(TumblingProcessingTimeWindows.of(Time.seconds(5))) //Flink 1.13 onward
                .sum(1) // Sum the number of words per partition
                .map(value -> value.f0 + "," + value.f1.toString() + "\n")
                .addSink(createSinkFromStaticConfig());

Compile the Application Code

To compile the application, do the following:

  1. Install Java and Maven if you haven't already. For more information, see Prerequisites in the Getting Started (DataStream API) tutorial.

  2. Compile the application with the following command: 

    mvn package -Dflink.version=1.15.3
    Note

    The provided source code relies on libraries from Java 11.

Compiling the application creates the application JAR file (target/aws-kinesis-analytics-java-apps-1.0.jar).

Upload the Apache Flink Streaming Java Code

In this section, you upload your application code to the Amazon S3 bucket that you created in the Create Dependent Resources section.

  1. In the Amazon S3 console, choose the ka-app-code-<username> bucket, and then choose Upload.

  2. In the Select files step, choose Add files. Navigate to the aws-kinesis-analytics-java-apps-1.0.jar file that you created in the previous step.

  3. You don't need to change any of the settings for the object, so choose Upload.

Your application code is now stored in an Amazon S3 bucket where your application can access it.

Create and Run the Managed Service for Apache Flink Application

Follow these steps to create, configure, update, and run the application using the console.

Create the Application

  1. Open the Managed Service for Apache Flink console at https://console.aws.amazon.com/flink

  2. On the Managed Service for Apache Flink dashboard, choose Create analytics application.

  3. On the Managed Service for Apache Flink - Create application page, provide the application details as follows:

    • For Application name, enter MyApplication.

    • For Runtime, choose Apache Flink.

    • Leave the version pulldown as Apache Flink version 1.15.2 (Recommended version).

  4. For Access permissions, choose Create / update IAM role kinesis-analytics-MyApplication-us-west-2.

  5. Choose Create application.

Note

When you create a Managed Service for Apache Flink application using the console, you have the option of having an IAM role and policy created for your application. Your application uses this role and policy to access its dependent resources. These IAM resources are named using your application name and Region as follows:

  • Policy: kinesis-analytics-service-MyApplication-us-west-2

  • Role: kinesisanalytics-MyApplication-us-west-2

Edit the IAM Policy

Edit the IAM policy to add permissions to access the Kinesis data streams.

  1. Open the IAM console at https://console.aws.amazon.com/iam/.

  2. Choose Policies. Choose the kinesis-analytics-service-MyApplication-us-west-2 policy that the console created for you in the previous section.

  3. On the Summary page, choose Edit policy. Choose the JSON tab.

  4. Add the highlighted section of the following policy example to the policy. Replace the sample account IDs (012345678901) with your account ID.

    {
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "ReadCode",
            "Effect": "Allow",
            "Action": [
                "s3:GetObject",
                "logs:DescribeLogGroups",
                "s3:GetObjectVersion"
            ],
            "Resource": [
                "arn:aws:logs:us-west-2:012345678901:log-group:*",
                "arn:aws:s3:::ka-app-code-<username>/aws-kinesis-analytics-java-apps-1.0.jar"
            ]
        },
        {
            "Sid": "DescribeLogStreams",
            "Effect": "Allow",
            "Action": "logs:DescribeLogStreams",
            "Resource": "arn:aws:logs:us-west-2:012345678901:log-group:/aws/kinesis-analytics/MyApplication:log-stream:*"
        },
        {
            "Sid": "PutLogEvents",
            "Effect": "Allow",
            "Action": "logs:PutLogEvents",
            "Resource": "arn:aws:logs:us-west-2:012345678901:log-group:/aws/kinesis-analytics/MyApplication:log-stream:kinesis-analytics-log-stream"
        },
        {
            "Sid": "ListCloudwatchLogGroups",
            "Effect": "Allow",
            "Action": [
                "logs:DescribeLogGroups"
            ],
            "Resource": [
                "arn:aws:logs:us-west-2:012345678901:log-group:*"
            ]
        },
        {
            "Sid": "ReadInputStream",
            "Effect": "Allow",
            "Action": "kinesis:*",
            "Resource": "arn:aws:kinesis:us-west-2:012345678901:stream/ExampleInputStream"
        },
        {
            "Sid": "WriteOutputStream",
            "Effect": "Allow",
            "Action": "kinesis:*",
            "Resource": "arn:aws:kinesis:us-west-2:012345678901:stream/ExampleOutputStream"
        }
    ]
}

Configure the Application

  1. On the MyApplication page, choose Configure.

  2. On the Configure application page, provide the Code location:

    • For Amazon S3 bucket, enter ka-app-code-<username>.

    • For Path to Amazon S3 object, enter aws-kinesis-analytics-java-apps-1.0.jar.

  3. Under Access to application resources, for Access permissions, choose Create / update IAM role kinesis-analytics-MyApplication-us-west-2.

  4. Under Monitoring, ensure that the Monitoring metrics level is set to Application.

  5. For CloudWatch logging, select the Enable check box.

  6. Choose Update.

Note

When you choose to enable Amazon CloudWatch logging, Managed Service for Apache Flink creates a log group and log stream for you. The names of these resources are as follows:

  • Log group: /aws/kinesis-analytics/MyApplication

  • Log stream: kinesis-analytics-log-stream

This log stream is used to monitor the application. This is not the same log stream that the application uses to send results.

Configure the Application Parallelism

This application example uses parallel execution of tasks. The following application code sets the parallelism of the min operator:

.setParallelism(3) // Set parallelism for the min operator

The application parallelism can't be greater than the provisioned parallelism, which has a default of 1. To increase your application's parallelism, use the following AWS CLI action:

aws kinesisanalyticsv2 update-application
      --application-name MyApplication
      --current-application-version-id <VersionId>
      --application-configuration-update "{\"FlinkApplicationConfigurationUpdate\": { \"ParallelismConfigurationUpdate\": {\"ParallelismUpdate\": 5, \"ConfigurationTypeUpdate\": \"CUSTOM\" }}}"

You can retrieve the current application version ID using the DescribeApplication or ListApplications actions.

Run the Application

The Flink job graph can be viewed by running the application, opening the Apache Flink dashboard, and choosing the desired Flink job.

You can check the Managed Service for Apache Flink metrics on the CloudWatch console to verify that the application is working.

Clean Up AWS Resources

This section includes procedures for cleaning up AWS resources created in the Sliding Window tutorial.

Delete Your Managed Service for Apache Flink Application

  1. Open the Managed Service for Apache Flink console at https://console.aws.amazon.com/flink

  2. In the Managed Service for Apache Flink panel, choose MyApplication.

  3. In the application's page, choose Delete and then confirm the deletion.

Delete Your Kinesis Data Streams

  1. Open the Kinesis console at https://console.aws.amazon.com/kinesis.

  2. In the Kinesis Data Streams panel, choose ExampleInputStream.

  3. In the ExampleInputStream page, choose Delete Kinesis Stream and then confirm the deletion.

  4. In the Kinesis streams page, choose the ExampleOutputStream, choose Actions, choose Delete, and then confirm the deletion.

Delete Your Amazon S3 Object and Bucket

  1. Open the Amazon S3 console at https://console.aws.amazon.com/s3/.

  2. Choose the ka-app-code-<username> bucket.

  3. Choose Delete and then enter the bucket name to confirm deletion.

Delete Your IAM Resources

  1. Open the IAM console at https://console.aws.amazon.com/iam/.

  2. In the navigation bar, choose Policies.

  3. In the filter control, enter kinesis.

  4. Choose the kinesis-analytics-service-MyApplication-us-west-2 policy.

  5. Choose Policy Actions and then choose Delete.

  6. In the navigation bar, choose Roles.

  7. Choose the kinesis-analytics-MyApplication-us-west-2 role.

  8. Choose Delete role and then confirm the deletion.

Delete Your CloudWatch Resources

  1. Open the CloudWatch console at https://console.aws.amazon.com/cloudwatch/.

  2. In the navigation bar, choose Logs.

  3. Choose the /aws/kinesis-analytics/MyApplication log group.

  4. Choose Delete Log Group and then confirm the deletion.