Monthly Archives: May 2019

Amazon Web Services Simple Queue Service (AWS SQS) Using the Java 2 Software Development Kit

Introduction

AWS SQS Message Queues are a way to exchange messages between applications. Senders, send data objects to a queue and receivers, receive objects from a queue. Amazon’s Simple Queue Service (AWS SQS) is a service offered by AWS that offers scalability and reliability by being distributed across Amazon.

A message queue decouples applications. An message producer only knows about the queue and knows nothing about the queue’s consumers. Likewise, a message consumer only knows about the queue and knows nothing about the queue’s other consumers or producers. Moreover, producers and consumers know nothing about timing, and are asynchronous.

For more on queues and message-passing in general, there are many resources online. Here is a good reference from MIT: Reading 22: Queues and Message-Passing.

Use Case

Suspend disbelief, or more accurately, simply build the system regardless of what you think about the soundness behind the business plan. Famous entrepreneur John Bunyan from Washington State has a plan to get rich and famous by finally proving conclusively that Bigfoot – or Sasquatch for the cultured – exists and uses the extensive system of hiking trails to move around.

Against his accountant’s advice, he liquidated half his fortune to install a series of hidden cameras along Washington State’s hiking trails to take photos every fifteen minutes. As he is a busy man, he does not have time to analyze all the photos personally, and so he want’s image analysis software to analyze the images. If the software registers a Sasquatch, he wants the images to personally go to his email account so he can register the image as a Squatch or not.

Now, with if 10,000 cameras take a picture every 15 minutes, that is 600,000 images per hour. Assume each image takes up to five minutes to process. Hopefully you can see, we have a scalability issue.

There are various ways to deal with this scalability issue, but as this is a tutorial on SQS, we use AWS SQS. And, as I am fond of admonishing in all my tutorials, if the “business case” seems suspect, then suspend disbelief and focus on the AWS code.

Design

Enough apologizing for the business case, let’s focus on the application’s design. The following diagram illustrates the dilemma.

  • Every n minutes a Station sends an observation to an AWS queue.
  • There are 1 or more SquatchFinder components who’s job is to pick up an observation from the queue and process the observation.
  • Station is the producer while SasquatchFinder is the consumer.
Stations send observations to the queue and SasquatchFinders get observations from the queue.
Queues implement an asynchronous Producer/Consumer design pattern.

We can formalize our requirements with a simple class diagram. A Station creates an Observation. A SasquatchFinder processes an Observation.

Class diagram illustrating the design.

All communication with AWS isfrom external processes is via its REST API. AWS SQS is no different. Moreover, SQS queues only accept textual data. But a common need is for the queue to accept binary data, such as an image. Also, JSON is a textual data transport format.

We can translate the Observation into a JSON document. The image is converted to base64 encoding so it can be represented as text. Note the encodedImage in this tutorial is always truncated with <snip>, as the base64 string is quite long.

{ 
  timestamp: “1558493503”,
  latitude:”46.879967”,
  longitude:”-121.726906”,
  encodedImage:"/9j/4AA <snip> 3QEUBGX/9k="
}

Base64 Encoding

Images are binary. However, all binary can be represented by a String provided it is encoded and decoded correctly. Base64 is an encoding scheme that is converts binary to a string. It’s useful because it allows embedding binary data, such as an image, in a textual file, such as a webpage or JSON document. AWS SQS queues only allow textual data, and so if you wish to store an image on an AWS SQS queue, you must convert it to a string. And the easiest way to accomplish this is by using Base64 format to encode binary data to strings when transporting data and decode strings to binary data when storing the data. For an example of Base64 and DynamoDB, refer to this site’s tutorial: Using the AWS DynamoDB Low-Level Java API – Sprint Boot Rest Application.

Station – Producer

Before coding the application, let’s create a queue. You can create a queue via the Java 2 API SDK; however, here we create the queue manually and then use this queue to send and receive messages.

Create SQSQueue

  • Navigate to the SQS console and select standard Queue.
  • Click the Configure Queue button.
  • Name the queue SasquatchImageQueue.
  • Accept the defaults for the Queue Attributes.
  • After creating the queue you should see a screen similar to the following.
  • Click on the Permissions tab and notice that we have not created a permission. We return to the Permissions tab after creating the two necessary users.

There are two types of queues offered by AWS SQS, Standard Queues and First In First Out (FIFO) Queues. Standard queues provide what is called best-effort ordering. Although messages are usually delivered in the order they are received, there are no guarantees. Moreover, messages can also be processed more than once. FIFO queues, in contrast, guarantee first in first out delivery and processing only once.

In this tutorial we primarily use standard queues. However, toward the end of this tutorial we illustrate using a FIFO queue.

Create SQSQueue Users

We need to create two users, one to interact with the queue for sending messages and another for receiving messages. If you have created IAM users before, note we do not assign the user to any group or assign any policies. Instead, we allow the queue to determine its permissions. Of course, we assign the user programmatic access and download the credentials file.

  • Navigate to the IAM console and create a new user called SasquatchProducerUser that has programmatic access.
  • Save the user’s credentials locally.
  • Create a second user called SasquatchConsumerUser that also has programmatic access.
  • Save the user’s credentials locally.
  • You should have two users created with programmatic access.

Queue Permissions

Initially only a queue’s creator, or owner, can read or write to a queue. The creator must grant permissions. We do this using a queue policy. We write the policy using the ASW SQS Console, although you write it manually if you wished.

Consumer Permissions

  • Navigate to the SasquatchConsumerUser summary screen and copy the Amazon Resource Name (ARN).

The ARN should appear similar to the following.

arn:aws:iam::743327341874:user/SasquatchConsumer

The Amazon Resource Number, or ARN, uniquely identifies an Amazon resource, in this case, the SasquatchConsumer user.

  • Return to the SQS console and select the SasquatchImageQueue and click on the Permissions tab.
  • Click Add a Permission.
  • In the resultant popup, paste the ARN in the Principal text box.
  • Check the DeleteMessage, GetQueueUrl, and ReceiveMessage Actions.
  • Click Save Changes.
  • After creating the SasquatchConsumerUser, navigate to the SasquatchProducerUser and copy the ARN for the producer.
arn:aws:iam::743327341874:user/SasquatchProducerUser
  • Navigate back to the SQS Queue and add this user to the queue as a permission. Allow the ChangeMessageVisibility, DeleteMessage, GetQueueAttributes, GetQueueUrl, PurgeQueue, and SendMessage Actions.
  • After adding the permissions for both users the queue should appear similar to the following image.

If you are still uncertain as to adding a permission to a queue, here is a tutorial by Amazon: Adding Permissions to an Amazon SQS Queue. You can also add Server-Side Encryption, as this tutorial illustrates: Creating an Amazon SQS Queue with Server-Side Encryption (SSE).

Although we do not discuss Policy documents, the following illustrates that a JSON document underlies the settings we set using the console. It is, however, important you understand policy documents, as they are at the heart of AWS security. For more information on AWS SQS Policies refer to this documentation: Using Identity-Based (IAM) Policies for Amazon SQS.

One thing to note is that here we assigned permissions to the queue using AWS SQS rather than the consumer or producer user we created. We could have just as easily used an IAM Policy, as the documentation in the link in the preceding paragraph discusses.

Sending Message Via Console

Although there is probably rarely a business reason, for testing purposes you can manually add a message to a queue. Although we will not use the message, let’s explore sending a message using the AWS SQS Console.

  • Refer to the observations.json document and copy one of the observations. Of course, in the code listing below the image is truncated.
    {
      "stationid": 221,
      "date": "1992-03-12",
      "time": "091312",
      "image": "/9j/4AA <snip> 0Wxkf/9k="
    }
  • Select the queue and from Queue Actions select Send a Message.
  • Copy a single message from observations.json and add the entry to the Message Body.
  • Click Send Message and within a minute the Messages Available column should show one message on the queue.
  • Purge the queue by selecting Purge Queue from Queue Actions.

Java Project – Producer

As discussed, a producer, well, produces messages. If we fully implemented the design above we would have many Stations and many . However, to keep the tutorial simple we limit ourselves to one Station in one project.

Project Setup

Although I developed the tutorial using Eclipse, you can use your own IDE or even the command-line. However, you really should use Maven or Gradle. Here we use Maven. It is assumed you are familiar with using Maven to build Java projects.

POM

  • Create a new project named SQSTutorialProducer.
  • Create or overwrite the POM file with the following POM.
<project xmlns="http://maven.apache.org/POM/4.0.0"
	xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
	xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
  <modelVersion>4.0.0</modelVersion>
  <groupId>com.tutorial.aws</groupId>
  <artifactId>SQSTutorialProducer</artifactId>
  <version>0.0.1-SNAPSHOT</version>
  <properties>
    <java.version>1.8</java.version>
    <maven.compiler.source>1.8</maven.compiler.source>
    <maven.compiler.target>1.8</maven.compiler.target>
  </properties>
  <dependencyManagement>
    <dependencies>
      <dependency>
	<groupId>software.amazon.awssdk</groupId>
	<artifactId>bom</artifactId>
	<version>2.5.25</version>
  	  <type>pom</type>
  	  <scope>import</scope>
	  </dependency>
    </dependencies>
  </dependencyManagement>
  <dependencies>
    <dependency>
      <artifactId>auth</artifactId>
      <groupId>software.amazon.awssdk</groupId>
    </dependency>
    <dependency>
      <artifactId>aws-core</artifactId>
      <groupId>software.amazon.awssdk</groupId>
    </dependency>
    <dependency>
      <groupId>software.amazon.awssdk</groupId>
      <artifactId>auth</artifactId>
    </dependency>
    <dependency>
      <artifactId>sqs</artifactId>
      <groupId>software.amazon.awssdk</groupId>
    </dependency>
   <dependency>
       <groupId>org.slf4j</groupId>
       <artifactId>slf4j-api</artifactId>
       <version>1.7.5</version>
   </dependency>
   <dependency>
       <groupId>org.slf4j</groupId>
       <artifactId>slf4j-simple</artifactId>
       <version>1.6.4</version>
   </dependency>
  </dependencies>
  <build>
    <plugins>
      <plugin>
        <groupId>org.apache.maven.plugins</groupId>
	<artifactId>maven-dependency-plugin</artifactId>
	<executions>
  	  <execution>
	    <id>copy-dependencies</id>
	    <phase>prepare-package</phase>
	    <goals>
	      <goal>copy-dependencies</goal>
	    </goals>
	    <configuration>				 <outputDirectory>${project.build.directory}/lib</outputDirectory>				 
              <overWriteReleases>false</overWriteReleases>
              <overWriteSnapshots>false</overWriteSnapshots>						 
              <overWriteIfNewer>true</overWriteIfNewer>
	    </configuration>
        </execution>
    </executions>
  </plugin>
  <plugin>
    <groupId>org.apache.maven.plugins</groupId>
    <artifactId>maven-jar-plugin</artifactId>
    <configuration>
      <archive>
        <manifest>
	  <addClasspath>true</addClasspath>
	  <classpathPrefix>lib/</classpathPrefix>
	  <mainClass>com.aws.tutorial.sqs.main.Station</mainClass>
	</manifest>
      </archive>
    </configuration>
  </plugin>
</plugins>
</build>
</project>

In the POM we use the AWS BOM so we can avoid specifying AWS library versions. We add dependencies for the required AWS libraries. We also specify that maven is to build an executable jar with the required dependencies packaged in the jar.

Notice the following.

  <properties>
    <java.version>1.8</java.version>
    <maven.compiler.source>1.8</maven.compiler.source>
    <maven.compiler.target>1.8</maven.compiler.target>
  </properties>

If we do not specify Java 1.8 or higher, the compilation will fail, as the AWS builders are static interface methods that do not work with older Java versions. Although on your machine, the code might compile, you could have issues if you have multiple Java SDKs on your computer. By explicitly setting the version, source, and target we avoid any potential issues with compilation.

Station

Let’s create a simple executable Java class named Station. This will simulate a bona-fide message producer.

  • First create an com.aws.tutorial.sqs.main package.
  • Create a class named Station with a main method in the created package.
  • Have the main method printout a message that the class executed.
package com.aws.tutorial.sqs.main;

public class Station {
  public static void main(String[] args) {
    System.out.println("Station running....");
  }
}

Executable Jar

  • Compile and package the project. If running from the command-line you would type the following.
$ mvn clean compile package
  • After building, execute the program from the command-line. The printout should appear.
$ java -jar SQSTutorialProducer-0.0.1-SNAPSHOT.jar 
Station running....

Now that we have created the consumer’s basic structure, we can modify it to send an SQS message.

Sending A Message

In this example we send a message to the queue using the SDK. The data payload is a string of JSON data. You use hardcoded data to send to the queue. Obviously in a real-world application the data would come from a different source. To simulate sending messages from a bona-fide producer, a delay is introduced between sending each message.

  • Before modifying the program, create a new class named TestData in the com.aws.tutorial.sqs.main package.
  • Copy three observations from the observations.json file.
  • Or, if you do not wish escaping the strings yourself, use the TestData.java from this tutorial’s Git project. Note: if you use Eclipse, it will escape the strings for you when you paste the string immediately after the opening quotation. The image’s base64 code is shortened so they can be easily displayed.
package com.aws.tutorial.sqs.main;

public class TestData {

	public static String observationOne = "    {\n" + 
			"      \"stationid\": 221,\n" + 
			"      \"date\": \"2019-03-12\",\n" + 
			"      \"time\": \"091312\",\n" + 
   		        "      \"image\": \"/9j/4A <snip> \"\n" + 
			"    }";
	
	public static String observationTwo = "    {\n" + 
			"      \"stationid\": 222,\n" + 
			"      \"date\": \"2016-02-09\",\n" + 
			"      \"time\": \"091312\",\n" + 
			"      \"image\": \"/9j/4A <snip> \"\n" +  
			"    }";
	
	public static String observationThree = "    {\n" + 
			"      \"stationid\": 223,\n" + 
			"      \"date\": \"2017-12-22\",\n" + 
			"      \"time\": \"091312\",\n" + 
			"      \"image\": \"/9j/4A <snip> \"\n" + 
			"    }";
}
  • Modify Station to have a constructor that takes three strings, the key, secret key, and the queue’s URL.
  • Create two member variables, one of type SqsClient and the other String.
  • In the Station constructor initialize the SqsClient.
  • Create a method named sendMessage that sends the message to the queue.
  • Finally, modify main to send all three messages in TestData.java and pause between sending each message.
package com.aws.tutorial.sqs.main;

import javax.annotation.PreDestroy;
import software.amazon.awssdk.auth.credentials.AwsBasicCredentials;
import software.amazon.awssdk.auth.credentials.StaticCredentialsProvider;
import software.amazon.awssdk.regions.Region;
import software.amazon.awssdk.services.sqs.SqsClient;
import software.amazon.awssdk.services.sqs.model.SendMessageRequest;
import software.amazon.awssdk.services.sqs.model.SendMessageResponse;

public class Station {

  SqsClient sqsClient;
  String queueUrl;
	
  public Station(String key, String secretKey, String queueUrl) {
    AwsBasicCredentials awsCreds = AwsBasicCredentials.create(key, secretKey);

    this.sqsClient = SqsClient.builder()
      .credentialsProvider(StaticCredentialsProvider
      .create(awsCreds)).region(Region.US_EAST_1).build();

    this.queueUrl = queueUrl;
  }
	
  public String sendMessage(String message) {
    SendMessageRequest request = SendMessageRequest.builder()
      .queueUrl(this.queueUrl).messageBody(message)
      .delaySeconds(5).build();		

    SendMessageResponse response = this.sqsClient.sendMessage(request);
    return response.messageId();
  }

  @PreDestroy
  public void preDestroy() {
    this.sqsClient.close();
  }

  public static void main(String[] args) {
    System.out.println("Station running....");
    Station station = new Station("AKIA22EODDUZONNX2EMP",
      "LUXJ5WQjW0p4bk1gC5oGBUi41rxA7oSvWWA/8SqH",
      "https://sqs.us-east-1.amazonaws.com/743327341874/SasquatchImageQueue");

    String id = station.sendMessage(TestData.observationOne);
    System.out.println("sent message: " + id);
    try {
      Thread.sleep(10000);
    } catch (InterruptedException e) {
        e.printStackTrace();
      }
    id = station.sendMessage(TestData.observationTwo);
    System.out.println("sent message: " + id);
    try {
      Thread.sleep(5000);
    } catch (InterruptedException e) {
      e.printStackTrace();
    }
    id = station.sendMessage(TestData.observationThree);
    System.out.println("sent message: " + id);
    }
}
  • Compile and run the application and you should see the following output.
Station running....
sent message: b861220e-a37a-424d-880c-5dd67a052967
sent message: 5185e68b-a16f-4300-8ee5-7ef5cca0eb53
sent message: 161f7444-ae7b-4890-b022-0447933054c3
  • Navigate to the queue in the AWS Console and you should see three messages in the Messages Available column.

The consumer has only one SqsClient instance that is initialized in the Station constructor and closed in a method annotated with the @PreDestroy annotation. This annotation is used to mark a method that should be called when a class is about to be destroyed for garbage collection.

Credentials

The client requires credentials to operate. This is the user account that the application uses to authenticate itself to the AWS SDK. Here we hardcode the credentials for simplicity. For more information on AWS Java 2 SDK and credentials, refer to SDK Documentation.

SqsClient

The SqsClient is an interface that extends SdkClient, and is the client for accessing AWS SQS service. You use the SqsClientBuilder to build the client. You build the client by passing the credentials and the region.

 this.sqsClient = SqsClient.builder()
      .credentialsProvider(StaticCredentialsProvider
      .create(awsCreds)).region(Region.US_EAST_1).build()

All requests to SQS must go through the client. Different types of requests are named accordingly. For instance requesting to send a message requires a SendMessageRequest, requesting to delete a message requires a DeleteMessageRequest. If you have worked with the other services offered by the Java 2 SDK such as DynamoDb or S3, then this pattern should be familiar.

SendMessageRequest

The SendMessageRequest wraps requests to send messages to the client. You build the request using a SendMessageRequestBuilder. Above we are setting the queue’s URL, the message’s body, and how long to delay before sending the message. We obtained the queue’s URL from the AWS SDK Console.

SendMessageRequest request = SendMessageRequest.builder()
  .queueUrl(this.queueUrl).messageBody(message)
  .delaySeconds(5).build();
The URL is in the Details tab of the queue in the AWS Console.

SendMessageResponse

The client sends the request and receives a response. The SendMessageResponse wraps the response. The method then returns the messageId and main prints the value to the console.

SendMessageResponse response = this.sqsClient.sendMessage(request);
return response.messageId();

Now that we have created three messages and sent them to SQS, we can write a consumer to consume the messages. Now let’s create a Java project named SQSTutorialConsumer.

Java Project – Consumer

Consumers, well, consume messages. Let’s create a consumer for the messages on the queue. As with the producer, we greatly simplify the consumer by creating an executable class that runs from the command-line.

Project Setup

Let’s create a Java Maven project for the Consumer.

POM

  • Create a Java project named SQSTutorialConsumer as a Maven project.
  • Create a POM file with the following.
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
  <modelVersion>4.0.0</modelVersion>
  <groupId>com.tutorial.aws</groupId>
  <artifactId>SQSTutorialConsumer</artifactId>
  <version>0.0.1-SNAPSHOT</version>
  <properties>
    <java.version>1.8</java.version>
    <maven.compiler.source>1.8</maven.compiler.source>
    <maven.compiler.target>1.8</maven.compiler.target>
  </properties>
  <dependencyManagement>
    <dependencies>
      <dependency>
	<groupId>software.amazon.awssdk</groupId>
	<artifactId>bom</artifactId>
	<version>2.5.25</version>
  	  <type>pom</type>
  	  <scope>import</scope>
	  </dependency>
    </dependencies>
  </dependencyManagement>
  <dependencies>
    <dependency>
      <artifactId>auth</artifactId>
      <groupId>software.amazon.awssdk</groupId>
    </dependency>
    <dependency>
      <artifactId>aws-core</artifactId>
      <groupId>software.amazon.awssdk</groupId>
    </dependency>
    <dependency>
      <groupId>software.amazon.awssdk</groupId>
      <artifactId>auth</artifactId>
    </dependency>
    <dependency>
      <artifactId>sqs</artifactId>
      <groupId>software.amazon.awssdk</groupId>
    </dependency>
   <dependency>
       <groupId>org.slf4j</groupId>
       <artifactId>slf4j-api</artifactId>
       <version>1.7.5</version>
   </dependency>
   <dependency>
       <groupId>org.slf4j</groupId>
       <artifactId>slf4j-simple</artifactId>
       <version>1.6.4</version>
   </dependency>
  </dependencies>
  <build>
    <plugins>
      <plugin>
        <groupId>org.apache.maven.plugins</groupId>
	<artifactId>maven-dependency-plugin</artifactId>
	<executions>
  	  <execution>
	    <id>copy-dependencies</id>
	    <phase>prepare-package</phase>
	    <goals>
	      <goal>copy-dependencies</goal>
	    </goals>
	    <configuration>				 <outputDirectory>${project.build.directory}/lib</outputDirectory>				 
              <overWriteReleases>false</overWriteReleases>
              <overWriteSnapshots>false</overWriteSnapshots>						 
              <overWriteIfNewer>true</overWriteIfNewer>
	    </configuration>
        </execution>
    </executions>
  </plugin>
  <plugin>
    <groupId>org.apache.maven.plugins</groupId>
    <artifactId>maven-jar-plugin</artifactId>
    <configuration>
      <archive>
        <manifest>
	  <addClasspath>true</addClasspath>
	  <classpathPrefix>lib/</classpathPrefix>
	  <mainClass>com.aws.tutorial.sqs.main.SasquatchFinder</mainClass>
	</manifest>
      </archive>
    </configuration>
  </plugin>
</plugins>
</build>
</project>

SasquatchFinder

  • First create a com.aws.tutorial.sqs.main package.
  • Next create a class named SasquatchFinder in the package.
  • Create a main method in the class and have it printout that it ran.
package com.aws.tutorial.sqs.main;

public class SasquatchFinder {
  public static void main(String[] args) {
    System.out.println("SasquatchFinder running....");
  }
}
  • Build the project.
$ mvn clean compile package
  • After building the project, execute the program from the command-line.
$ java -jar SQSTutorialConsumer-0.0.1-SNAPSHOT.jar 
SasquatchFinder running....

Now that we have the project’s basic outline, we can add code to receive messages.

Receive Message

  • As with the Station in the SQSTutorialProducer project, create member variables.
  • Create a main method that initializes the SqsClient. Be certain to use the consumer’s credentials and not the producer’s.
  • Create a new method named processMessage and have it use a ReceiveMessageRequest to receive a message.
  • Create a new method named deleteMessage and have it use a DeleteMessageRequest to delete a message.
  • Modify processMessage to call deleteMessage after a delay.
  • Modify main to loop continuously processing messages.
package com.aws.tutorial.sqs.main;

import java.util.List;
import software.amazon.awssdk.auth.credentials.AwsBasicCredentials;
import software.amazon.awssdk.auth.credentials.StaticCredentialsProvider;
import software.amazon.awssdk.regions.Region;
import software.amazon.awssdk.services.sqs.SqsClient;
import software.amazon.awssdk.services.sqs.model.DeleteMessageRequest;
import software.amazon.awssdk.services.sqs.model.Message;
import software.amazon.awssdk.services.sqs.model.ReceiveMessageRequest;

public class SasquatchFinder {

	private SqsClient sqsClient;
	private String queueUrl;
	public static int finderId = 1;

	public SasquatchFinder(String key, String secretKey, String queueUrl) {
		AwsBasicCredentials awsCreds = AwsBasicCredentials.create(key, secretKey);
		this.sqsClient = SqsClient.builder().credentialsProvider(StaticCredentialsProvider.create(awsCreds))
				.region(Region.US_EAST_1).build();
		this.queueUrl = queueUrl;
	}

	public void processMessage() {

		ReceiveMessageRequest receiveMessageRequest = ReceiveMessageRequest.builder().queueUrl(this.queueUrl)
				.maxNumberOfMessages(1).build();
		List<Message> messages = this.sqsClient.receiveMessage(receiveMessageRequest).messages();
		if(messages == null || messages.size() == 0) return;
		messages.stream().map(s -> s.body()).forEach(System.out::println);
		try {
			System.out.println("sleeping for 10 seconds...");
			Thread.sleep(10000);
		  this.deleteMessage(messages);
		} catch (InterruptedException e) {
			e.printStackTrace();
		}
	}

	public void deleteMessage(List<Message> messages) {
		String receiptHandle = messages.get(0).receiptHandle();
		DeleteMessageRequest deleteRequest = DeleteMessageRequest.builder().queueUrl(this.queueUrl)
				.receiptHandle(receiptHandle).build();
		this.sqsClient.deleteMessage(deleteRequest);
	}

	public static void main(String[] args) {
		System.out.println("SasquatchFinder " + SasquatchFinder.finderId + " running....");
		SasquatchFinder finder = new SasquatchFinder("AKIA22EODDUZAMDPWSX7", "805hbufO3Sn18eDsBDrOzCgB/eT5KVPM/AIkIpoZ",
				"https://sqs.us-east-1.amazonaws.com/743327341874/SasquatchImageQueue");
		try {
			while (true) {
				finder.processMessage();
			}
		} catch (Exception e) {
			e.printStackTrace();
		}
		System.out.println("SasquatchFinder " + SasquatchFinder.finderId + " stopped.");
	}
}
  • Compile and run the producer and, if you had ran the consumer in the previous section, you should see the following output.
SasquatchFinder 1 running....
    {
      "stationid": 221,
      "date": "2019-03-12",
      "time": "091312",
      "image": "/9j/4AAQ <snip> kf/9k="
    }
sleeping for 10 seconds...
    {
      "stationid": 223,
      "date": "2017-12-22",
      "time": "091312",
      "image": "/9j/4AAQ <snip> kf/9k="
    }
sleeping for 10 seconds...
    {
      "stationid": 222,
      "date": "2016-02-09",
      "time": "091312",
      "image": "/9j/4AAQ <snip> kf/9k="
    }
sleeping for 10 seconds...
  • Navigate to the queue in the AWS Console and you should see no messages, as they were deleted after processing.

In this simple consumer we first create a client for interacting with the queue. We then obtain a single message from the queue. The program pauses to simulate processing. It then deletes the message from the queue by using the receiptHandle.

Because the program loops, it processes all three messages place on the queue when we created the consumer.

ReceiveMessageRequest

The ReceiveMessageRequest wraps the request to receive a message from an SQS queue. We use a builder to create the request. Then we specify the queue URL and the maximum number of messages to fetch. Finally, we specified a single message; however, you can specify multiple messages if desired.

ReceiveMessageRequest receiveMessageRequest = ReceiveMessageRequest.builder().queueUrl(this.queueUrl)
				.maxNumberOfMessages(1).build();

DeleteMessageRequest

After processing the message you should delete it from the queue. We do this by obtaining the receiptHandle of the received message which is then used to delete the message.

String receiptHandle = messages.get(0).receiptHandle();
		DeleteMessageRequest deleteRequest = DeleteMessageRequest.builder().queueUrl(this.queueUrl)
				.receiptHandle(receiptHandle).build();

The program processes all messages on the queue. This is a simple consumer, but you could have multiple consumers consuming messages from the same queue.

Message Visibility

A message might be processed twice when using a standard queue. A message, when picked up by a consumer for processing becomes invisible for a configurable time. When we created the queue we accepted the visibility timeout of 30 seconds. However, if processing takes longer than the visibility timeout, the message can be processed by another consumer. The following diagram illustrates.

There is a following wrinkle. What happens when the message is deleted from the queue a second time?

  • Open the SQS Console and send a single message to the queue.
  • Modify SasquatchFinder to sleep for 40 seconds between each message.
public void processMessage() {
  ReceiveMessageRequest receiveMessageRequest = ReceiveMessageRequest
    .builder().queueUrl(this.queueUrl).maxNumberOfMessages(1).build();

  List<Message> messages = this.sqsClient
    .receiveMessage(receiveMessageRequest).messages();

  if(messages == null || messages.size() == 0){
   return;
 }
  messages.stream().map(s -> s.body()).forEach(System.out::println);
  try {
    System.out.println("sleeping for 40 seconds...");
    Thread.sleep(40000);
    this.deleteMessage(messages);
  } catch (InterruptedException e) {
      e.printStackTrace();
    }
}
  • After building the application, open two command-line windows and execute the program in the two different windows at the same time.

One running instance gets the message from the queue. The message’s visibility timeout set at 30 seconds begins. The instance sleeps for 40 seconds to simulate processing.

$ java -jar SQSTutorialConsumer-0.0.1-SNAPSHOT.jar 2
SasquatchFinder 2 running....
mymessage
sleeping for 40 seconds...

The other instance finds no message on the queue, as the message is not visible.

$ java -jar SQSTutorialConsumer-0.0.1-SNAPSHOT.jar 1
SasquatchFinder 1 running....

However, after thirty seconds the message is visible again on the queue and it is picked up and processed by the other instance.

$ java -jar SQSTutorialConsumer-0.0.1-SNAPSHOT.jar 1
SasquatchFinder 1 running....
mymessage
sleeping for 40 seconds...

Meanwhile, the instance that first picked up the message finishes processing and deletes the message. In reality, it attempts to delete the message. But, as the other process already requested the message and a new receipt handle was issued, the message is not truly deleted.

$ java -jar SQSTutorialConsumer-0.0.1-SNAPSHOT.jar 2
SasquatchFinder 2 running....
mymessage
sleeping for 40 seconds...
Deleted message AQEB3/lhW8cY2cTsl2gd/GOsyPrt1J/SQn+ZR06ngf24aL5C8SqfUSPZfAl4uc2IwuZuLhZ/5BXYLWVU7AvmgSf0kb4zm0owKh01EXC4pGhvtNSsioLnk3nd4KiS5YEUO/EssCnRM1we7rXw0eLyd2LehOpPOZ49893lIJ6opy1vamQxxk6C+7iGcWbY0dMNTvrZqVaZw2JW/eZV5wI99rdUwRP16+RFj7XWsxEI5KJcExgnWY3jDRQv1mXqe5ZgWI9M7mqPH/rrx8afBdV2P53B7OK0uRm3vUGMzmW/xUgbsxsy5UB0+DZGLaccUAbegtC74LQ6BLZs64VlFxc8jAC2sp2gheLAZ849j4JkMrA8nWf+P+xKCjqdALeGrN754DcxnvhZv79R6sOGcp2lBtTOsA== by SasquatchFinder 2

As the message is still being processed by the second instance, the first does not see the message. The second instance then deletes the message.

$ java -jar SQSTutorialConsumer-0.0.1-SNAPSHOT.jar 1
SasquatchFinder 1 running....
mymessage
sleeping for 40 seconds...
Deleted message AQEBgZK7kq12asCcVVNbFQNlQmmgYTXXO8OzgoJzHpAnqdsBtMnaBxSBvjjgyVqO3nqYwuhFoxPWgXhUoUcgDzejHHSG6dM/VNG1Wdv3Q93THsJPj6BSQSH/sLjX7qvdFYT20Es0jdhN4dQTNMPyaA3sA7a2x025cUYLsegKfMlWVfCDThABbn+0evwgkn3hmzwLBvAWZEGIp0mooZvYf6WiLcblbqCnx+Gh5j5/XvmIpWuT9ux3DQSTYH+f+XdfUxclXP6exwAYyyFm7xHJnlF9LXcRcKmv2QitpQjgjK3yQBLrogU6dPf8Zp34K8iwMr1TBXEi5mZnfPSA7Cl3a4N2c+MxB+OupGIGGY6uoy2gFLSiaaunsij/weB0FFaYaE/MFhMsXdMMhNho2o/lrq6SOA== by SasquatchFinder 1

Notice that both messages have a different receiptHandle. The queue has an internal mechanism to avoid errors when a message is processed and subsequently deleted twice. However, it does not prevent processing a message multiple times. If we manipulated the processing time and/or the visibility timeout, we could have the message processed even more times.

To actually delete the underlying message, the most recent receipt handle must be provided. So in our example above, the first attempt to delete the message came after the second receipt handle was returned and so the message was not deleted. But the second attempt to delete the message was the most recent receipt handle and so the message was deleted. To delete a message you must pass the most recently issued receipt handle.

You should design your system to not be dependent upon the number of times a message is processed. Your system should be idempotent. If you need strict processing of once and only once, then use a FIFO queue.

Message Attributes & Dead Letter Queue

Let’s explore two topics important when working with AWS SQS queues: message attributes and dead letter queues. A message can have associated metadata. However, to receive messages with associated metadata the ReceiveMessageRequest must be explicitly instructed to fetch the associated metadata in addition to the message itself. A message might not be successfully processed. Rather than leaving the message on the queue to fail indefinitely, a dead letter queue can be configured to send message that fail a configurable number of times.

DeadLetter Queue

  • Create a new standard queue named DeadLetterQueue.
  • SelectSasquatchImageQueue and from the Queue Actions dropdown select Configure Queue.
  • Modify SasquatchImageQueue to use DeadLetterQueue for its Dead Letter Queue.

Message Attributes

  • Select SasquatchImageQueue and send a new message.
  • When creating the message, add two message attributes.
  • Open the SQSTutorialConsumer project and modify the processMessage method in SasquatchFinder. Note that you comment the call to delete the message.
public void processMessage() {
  ReceiveMessageRequest receiveMessageRequest = 
    ReceiveMessageRequest.builder().queueUrl(this.queueUrl)	 		 	 
     .maxNumberOfMessages(1).messageAttributeNames("*").build();		

List<Message> messages = 
  this.sqsClient.receiveMessage(receiveMessageRequest)
    .messages();

  if (messages == null || messages.size() == 0) {			
    return;		
  }

  messages.stream().map(s -> s.body()).forEach(System.out::println);

  for (Message message : messages) {
    System.out.println(message.messageId());
    Map<String, MessageAttributeValue> attributes = message
      .messageAttributes();

    Set<String> keys = attributes.keySet();
    
    for (String key : keys) {
      System.out.println(key + ":" + attributes.get(key).stringValue());
    }
  }
  try {
    System.out.println("sleeping for 10 seconds...");
    Thread.sleep(10000);
    //this.deleteMessage(messages);
  } catch (InterruptedException e) {
    e.printStackTrace();
  }
}
  • Compile and run the application. The message should process three times.
SasquatchFinder 1 running....
abc
e6ede972-9a6d-4c86-8c00-b16fe18977ff
attribute1:abc
attribute2:ded
sleeping for 10 seconds...
abc
e6ede972-9a6d-4c86-8c00-b16fe18977ff
attribute1:abc
attribute2:ded
sleeping for 10 seconds...
abc
e6ede972-9a6d-4c86-8c00-b16fe18977ff
attribute1:abc
attribute2:ded
sleeping for 10 seconds...
  • Return to the AWS Console and you should see that the message is placed on DeadLetterQueue.

To receive message attributes we were required to build the ReceiveMessageRequest with the explicit instruction to receive the message attributes by specifying messageAttributeNames. That method can take one or more attribute names, or a * to signify all attributes.

The message was sent to DeadLetterQueue, the queue configured as the SasquatchImageQueue dead letter queue.

If you wish to learn more about message attributes, here is a tutorial on Amazon’s website: Sending a Message with Attributes to an Amazon SQS Queue.

If you wish to learn more about dead-letter queues, here is a tutorial on Amazon’s website: Configuring an Amazon SQS Dead-Letter Queue.

maxNumberOfMessages

The ReceiveMessageRequest can receive more than one message at a time if more are available on a queue. Above we set the maximum number of messages as one. Let’s explore what happens we change the setting to more messages.

  • Modify the SasquatchFinder class by creating a new method called deleteMessages.
  • Have the method iterate over all received messages.
public void deleteMessages(List<Message> messages) {
  for(Message message:messages) {
    String receiptHandle = message.receiptHandle();
    DeleteMessageRequest deleteRequest = 
      DeleteMessageRequest.builder().queueUrl(this.queueUrl)
      .receiptHandle(receiptHandle).build();

    this.sqsClient.deleteMessage(deleteRequest);
    System.out.println("Deleted message " + receiptHandle 
    + " by SasquatchFinder " + SasquatchFinder.finderId);
  }
}
  • Modify processMessage to call deleteMessages rather than deleteMessage.
public void processMessage() {
  ReceiveMessageRequest receiveMessageRequest = 
    ReceiveMessageRequest.builder().queueUrl(this.queueUrl)
      .maxNumberOfMessages(10).messageAttributeNames("*").build();

  List<Message> messages = this.sqsClient
    .receiveMessage(receiveMessageRequest).messages();
  if (messages == null || messages.size() == 0) {
    return;
  }
  messages.stream().map(s -> s.body()).forEach(System.out::println);
  for (Message message : messages) {
    System.out.println(message.messageId());
    Map<String, MessageAttributeValue> attributes = message
      .messageAttributes();

    Set<String> keys = attributes.keySet();
    for (String key : keys) {
      System.out.println(key + ":" + attributes.get(key).stringValue());
    }
  }
  try {
    System.out.println("sleeping for 10 seconds...");
    Thread.sleep(10000);
    this.deleteMessages(messages);
  } catch (InterruptedException e) {
      e.printStackTrace();
    }
}
  • Compile the application.
  • After compiling, navigate to the AWS SQS Console and add five messages to the queue, with the message body of a1, a2, a3, a4, and a5 respectively.
  • Run the application and you should see output similar to the following.
SasquatchFinder 1 running....
a4
98a42736-e4b5-4dfd-9428-3e32d2ea145d
sleeping for 10 seconds...
Deleted message AQEBqmAqpGs85ERM2Y8EnD4zjBPO1KxomlhJgQCPQ+JO3gjYhRcZbflS1gKJT1kas0JId7bX4X+OmFWQfC8r+gZGr02jwBcKlhvSUIv0tx13Q88EPpzMJDNbB9w9oKbgR+hc8c0nZQPPjJ2uHu7KeQfTmIdK/dt49cs/GHFRZeq3pIUWN2jJO8h0UdlpLeFKbB96WjPvakAnXDFd46meejQvBod0x18L1Y1dBt6cZc5+9AbB6eb4bJjV5dKvyDCtIUP2XFZ8iwtZF1lxntzqXxdMGYCjzaQ/oqQ5EmVJ/pFMTgWlUTks+qVFMu7a/sOCfQm7bFwE3AofXQROAK3B0crssZTbzoqQ9oJv+nj0kn596gidN+gygrISvF9vESIG1M5Ll+Lk2ADWQeO+2UA/AJax3A== by SasquatchFinder 1
a1
a5
c167bb7a-f356-4d5b-aa0f-ea90075cef50
f0d79263-05da-485e-bf6a-fa6b3f9fe92a
sleeping for 10 seconds...
Deleted message AQEBGwtlQPM080KnHDAOWUsZKUQ4PWfLP2g/AFn0sr9ERDOJFssjl7rNXl3mL6ryqoH9EgiPEGyGXwPm6n/FSsfbPA9OSMJYLq0Fho9qtpkcoI0mmAqRPQ/7h0J++zAmmf3bflcD9BqJS+hz4a/Di8Eo6GB0oWJUFZEFYcKWnIUGMNgnQfY3xs1DF9UuNZdsu7h3KN9hGGy3vSTuLvJJox7DDHSgY+QU3nisT5dTSfltKc9vJMQq2mPxB/f2EUmgwKQ82f10A6lPlSjVuiyNtGkKVau3BorKINz3dtG+xAHd5wWfALFExyip7zFZl6wVsnzfKox9QBaxRSrukIfx3+w5rIilq1QujPpNqLKItlxOvaXvDvxi/8lWv31S5UNlY7ooEOYSIkh1wnNwXKY7ZP4aQQ== by SasquatchFinder 1
Deleted message AQEBLIUJqmODdigrnQ88hzta9Zr+PaQnctLqmYrQT0iU5ZxvaLPy0PGNTe7eKwLHbBvc+WdDbLXK951WaPYWoY9dbMJZMyRNnjEj3doGoUkmBOm0LzTs1xDkV+QPb3fGH3s+mxh2TFhX3KFOwXrvf4uqkpx9mHdGioMWa86NSsCUUEQ3vXGUXprSdGsSqXUsoAug7v6wBU3QIPzeQm8pRLmjbZPdx+ndeV80FwnFkxDfNx/mtpAibum4ON4CxDUB66jLC7nVRe0XxXBllM2G/brS7jseqbz+Q61qbFjLNWKo96kTBIrYDjvZEmcSQdp37cYMf4rO/vsr+/XCNUtbtcD8h9Xk8Fc+atcIsuQSlrLbYMplVgN3EwogYlXJsB9GSOlVQVpO+gwOLBXonXJ6i3EAbQ== by SasquatchFinder 1
a2
a5
e65fbcc2-2c4a-42f6-8b61-ca97dad4826e
b2bc665c-4c1c-42c7-b3d2-c1d5bf048ee9
sleeping for 10 seconds...
Deleted message AQEB2FZyDGQEOUgLxR9wIxAiJbk++Ktec9RLon3nAZr7bPeQu2QJ8iVxRMNg92ZgvoPY5qsBndcRGEQjI5zKHQ/r62tg4+LMWwFLSDBhDF3d55w6OosgLf+K7AIBICGAeTJanTkhCzQlWYM+HCDFEve+NhPsr5+/zabaeZrkKwSBh8E2jTCmr29LmNR6ld9Bz0NSboj5gi+Gxa3dTu+xPGMLMjANVQ1Qa1BhoYEI0QP8kl9gL8aBpLhkeW1eWXgRaRtRcTAVpjxF73ZlUEFVNyYeE/Mwz9ZT2lWRftj6dv5p2PUG5Z6VtbbBw/9AXQElJUTgfHKGd4iGEjo4A3l6ff6g/NVJzm/LkGq6909txbTIk8PSp5istS4bM318W6VG2ten9jYSU7+pj8H809AHoW3VEw== by SasquatchFinder 1
Deleted message AQEBMdzd33/uz7jNQMnBJu1ne7GRh9g2xHx6X0cPWLsU0emEN0G5SGbr3nF/9QklDrrW42BX1HW6IDWxvhlI4/bOByZobYOfjmv5Cr8rDEJYnNKWxqxBZeQqjArKTy90WeEs0puUw4l6PouEZOv35daHO0h01A8Dpk/oMlVBi/OZFCIM4fetG2tUxwa7eU15WiEF4mklZqqJx2bVTbdiZqwhOucgqXlyXK3IJ5FtBFd6ACtEyX1tQmIBn6njmk/CBuX0v5+LzaxlntHy9Q+FpjuPLEyyE5wGqIk9B8Kcqv469pnaE3UJJaCK7DxgG70rF/7M1kYzaDRbRBYJB9jS3W9b8qZpj1JU4JM4euH9xBP4j59MvdwgIs4lSPvO1F3NtdCuNeOOMF15/n1WvU2U31jSeg== by SasquatchFinder 1

As the example illustrates, you can specify the maximum number of messages to process, but not the number of messages. This should seem reasonable, as the consumer does not know how many messages are in the queue before processing. As an aside, note that the messages were not processed in the same order they were received in the listing above.

First In First Out (FIFO) Queue

Let’s modify the project to use a FIFO queue and rerun the two consumer instances simultaneously. Note that neither the consumer nor the producer know they queue’s type. They only know it’s url.

  • Create a new queue named SasquatchImageQueue.fifo of type FIFO Queue.
  • Click Quick-Create Queue.
  • Create a new permission, but let’s be lazy and check the Everybody checkbox and the All SQS Actions checkbox. You would obviously not do this in production.
  • Modify both the consumer and producer to use this queue’s URL.
https://sqs.us-east-1.amazonaws.com/743327341874/SasquatchImageQueue.fifo
  • Modify the sendMessage method in the producer. Note the removal of the delaySeconds and the addition of the messageGroupId.
public String sendMessage(String message) {	
  SendMessageRequest request = SendMessageRequest.builder()
      .queueUrl(this.queueUrl).messageBody(message)
      .messageGroupId("mygroup").build();

  SendMessageResponse response = this.sqsClient.sendMessage(request);
  return response.messageId();
}
  • Compile and run the producer application after changing the queue and the three messages are sent to the queue.
  • Compile and run the consumer application and the three messages are processed in the same order they were received.

Message Visibility

  • Modify SasquatchFinder processMessage to simulate processing by sleeping for 40 seconds.
public void processMessage() {
  ReceiveMessageRequest receiveMessageRequest = 
    ReceiveMessageRequest.builder().queueUrl(this.queueUrl)
    .maxNumberOfMessages(1).build();

  List<Message> messages = this.sqsClient
    .receiveMessage(receiveMessageRequest).messages();

  if(messages == null || messages.size() == 0) {
    return;
  }

  messages.stream().map(s -> s.body()).forEach(System.out::println);
  try {
	System.out.println("sleeping for 40 seconds...");
	Thread.sleep(40000);
        this.deleteMessage(messages);
  } catch (InterruptedException e) {
        e.printStackTrace();
  }
}
  • Compile and run the application. Note you get an SqsException.
SasquatchFinder 2 running....
messageMine
sleeping for 40 seconds...
software.amazon.awssdk.services.sqs.model.SqsException: Value AQEBBJL+BlwyhRLnQGxaIKDkkrEv1sU6VnHzYM51Q0UFdx2lDyWvKoI/JYcs7MktVJ1Nmyr1mCVX/cpcqS9dMqq7Ual92VLEXDS9hEYM/qg1vdEGHB60OktMzpidyWBenQQyybzXofO+pAdKOYpC/wiEw8GBPsmFDCHpVn1hxHeLSNJyw10SwNv3DTXQXk4Pe+v3yGf23bf8sDk7Rx7ApqWYi8n8z9uijZAQBdwuFpUrZslivMWCzid6AFOXI/k83+/tKnSMyT0/Mx0rng0v1k4WliSgv5YJo5HyEZTt+cOBwfA= for parameter ReceiptHandle is invalid. Reason: The receipt handle has expired. (Service: Sqs, Status Code: 400, Request ID: 845b9538-4104-5428-aa2f-c05092244385)
	at software.amazon.awssdk.core.internal.http.pipeline.stages.HandleResponseStage.handl <snip> at com.aws.tutorial.sqs.main.SasquatchFinder.main(SasquatchFinder.java:58)
SasquatchFinder 2 stopped.

Attempting to delete messages fail when executed after the visibility timeout window if using FIFO queues.

Conclusions

In this tutorial we created an Amazon SQS Queue. After creating the queue, we created a message producer and a message consumer using the AWS Java 2 SDK. We then explored several topics such as message attributes, dead-letter queues, and message visibility. We also created a FIFO queue.

Amazon’s SQS Queue is a easy to use queue that takes the infrastructure management hassle away from the organization. In this tutorial we only examined SQS basics. For more information, refer to both the Java 2 SDK Developer’s Guide and the SQS Developer’s Guide. Remember, the API from version 1 to 2 changed, so when in doubt, assume you need a builder for an object and that you must configure the object when building it. However, the API is consistent and once you start working with the API translating 1.1. code to 2 is intuitive.

GitHub Project

The GitHub Project, SQSTutorial is available here.

Using the AWS DynamoDB Java API – Spring Boot Rest Application

Introduction

In this tutorial we use the Amazon Web Services Java 2 Application Programming Interface (API) to create a Rest application using Spring Boot that reads and writes to a DynamoDB database. This tutorial assumes AWS familiarity, Java programming experience, and Spring Boot experience. However, even without this experience, this tutorial should still prove useful, as it provides considerable supplementary resources for you to review. If you want to learn the AWS DynamoDB Java API then this tutorial is for you.

Here we create a simple database consisting of “observation stations” and “observations” gathered via a camera. Whatever…suspend disbelief and just go with it. Now, suppose, the stations require a means of uploading observations to an associated DynamoDB table. We decide upon a Rest API for stations to upload data. We implement this API using a Spring Boot Rest application. Again, if this all sounds suspect, suspend disbelief and focus on the AWS code and not the application design.

In this tutorial we,

  • create two database tables using the DynamoDB console,
  • create a couple items using the console,
  • create an IAM programatic user,
  • create a Spring Boot application that provides Rest endpoints so a client application can,
    • write an observation,
    • read an observation,
    • update an observation,
    • delete an observation,
    • batch write multiple observations,
    • conditionally query for a station’s observations,
    • and conditionally update observations,
  • and test the Rest endpoints using Postman.

This tutorial’s purpose is to explore the DynamoDB, not introduce Spring Boot, Rest, or JSON and assumes basic knowledge of all three. However, if new to any of these topics, links are provided to learn them before continuing.

NoSQL Databases

DynamoDB is a key-value and document NoSQL database. If unfamiliar with NoSQL Document databases, you should familiarize yourself before continuing. The following is an introductory video introducing NoSQL Databases.

The following are two good written introductory articles covering NoSQL and DynamoDB.

https://youtu.be/ujWV3-m1pLo

Note that Amazon also offers DocumentDB, which we could use as an alternative to DynamoDB. However, DocumentDB will be covered in a different tutorial.

A DynamoDB database can be described as the following. Tables consist of items. An item has one or more attributes. In a table you define the partition key and optionally define a sort key. The partition key is a key-value pair that not only uniquely identifies an item, it determines how the item is distributed on a computer’s storage. A sort key not only logically sorts items, it stores the items accordingly. Obviously, there is more to NoSQL physical storage and how it achieves its scalability, but that is beyond this tutorial’s scope.

Amazon Web Services & DynamoDB

Amazon DynamoDB is a NoSQL key-value and document database offered as a cloud service. It is fully managed and allows users to avoid the administrative tasks associated with hosting an enterprise database. As with almost all Amazon’s offerings, it is accessible via a Rest API.

Amazon offers software development kits (SDKs) to simplify working with the Rest API. The languages offered are Java, C++, C#, Ruby, Python, JavaScript, NodeJs, PHP, Objective-C, and Go. In this article we use the Java API. There are currently two versions of the API, in this tutorial we use the Java 2 SDK.

The Java 2 AWS SDK is a rewrite of the Java 1.1 AWS SDK and changes from a more traditional programming paradigm of instantiating objects using constructors and then setting properties using setters to a fluent interface/builder programming style.

Fluent Interface

The fluent interface is a term created by Martin Fowler and Eric Evans. It refers to an programming style where the public methods (the API) can be chained together to perform a task. It is used by the AWS Java SDK 2.0 when using builders. The builder tasks perform tasks but then return an instance of the builder. This allows chaining methods together. For more information on the fluid interface and builders, refer to this blog post: Another builder pattern for Java.

DynamoDB Low-Level API

As with all AWS APIs, DynamoDB is exposed via Rest endpoints. The AWS SDKs provide an abstraction layer freeing you from calling Rest directly. Above that layer, the Java SDK provides a class named DynamoDBMapper that allows working with DynamoDB similarly to the Java Persistence Framework (JPA). Although useful, using the lower-level API is not that difficult. Moreover, there are many situations where you would not wish to create a dependency in your object model that relies on DynamoDB.

For example, suppose we implemented a system that stored widgets in DynamoDB. If using the DynamoDBMapper, the Widget model class would be dependent upon DynamoDB via annotations mapping the class to the Widgets table.

Alternatively, if we do not wish to use the DynamoDBMapper we can implement something similar to the following diagram. It is a typical DAO pattern, where the only direct dependency upon the AWS SDK is the WidgetDaoImpl class. For more information on the DAO design pattern, refer to the following introductory article: DAO Design Pattern.

In this tutorial on the AWS DynamoDB Java APl, we use the SDKs direct calls to the underlying DynamoDB Rest endpoints. As an aside, note that we do not use the DAO design pattern, instead putting the data access logic directly in the controller class for brevity. We do, however, use the Spring MVC design pattern using Rest.

  • Note: If willing to make your application dependent upon DynamoDBMapper, you should consider using it, as it greatly simplifies working with DynamoDB.
  • Java: DynamoDBMapper documentation

Tutorial Use-Case – Station Observations

Imagine we have stations responsible for taking photo observations. A station has a coordinate, address, and a name. A station has one Coordinate. A station has one address. A station can have unlimited observations.

Although this tutorial does not discuss NoSQL database design, from the diagram below it seems reasonable we need two tables, Station and Observation. Moreover, as the Observation table is very write intensive – stations will be sending observations to the application on a continuous basis – it makes sense to not include observations as a collection within a Station instance but keep it as a separate table. Remember, these are JSON documents, not relational tables. It is unreasonable to design Observations as a list of items within a Station and would lead to an excessively large and unwieldy database.

If there were enough Stations, for even more efficiency we might create a separate table for each station’s observations. This would allow greater throughput for both writing and reading observations. But, in this tutorial we simply define a stationid to identify an observation’s station and will create an index on this value.

DynamoDB Console

The AWS Management Console provides an easy web-based way of working with Amazon’s cloud services. Although not covered in this tutorial, for those new to AWS, here is a short video by Amazon explaining the Management Console. Note that AWS also offers a command-line interface and Application Programming Interfaces (APIs) for accessing its cloud services.

  • AWS Essentials: How to Navigate the AWS Console by LinuxAcademy.

Before beginning the programming portion of this tutorial we must create the DynamoDB database.

Create Station Table

  • After entering the AWS Management Console, navigate to the DynamoDB console.
  • Click the Create table button.
  • Provide Station as the table’s name and id as the table’s primary key.

Creating Station Items

Remember, DynamoDB is schema-less. We create an item but do not define a table’s schema. Instead, we create a couple items with the desired structure.

  • Click the Items tab and click the Create Item button.
  • Create an id and name attribute, assigning id as a Number datatype and name as a String. Assign the values 221 and “Potomac Falls” respectively.
  • Create an attribute named address and assign it the Map datatype.
  • Add a city, street, and zip attribute as String datatypes to the address map. In the example below, I assigned Potomac, 230 Falls Street, and 22333 as the attribute values.
  • Create coordinate as a Map and assign it a latitude and longitude attribute as Number datatypes. I assigned the values 38.993465 and -77.249247 as the latitude and longitude values.
  • Repeat for one more station.

We created two items in the Station table. Here are the two items as JSON.

{
  "address": {
    "city": "Potomac",
    "street": "230 Falls Street",
    "zip": "22333"
  },
  "coordinate": {
    "latitude": 38.993465,
    "longitude": -77.249247
  },
  "id": 221,
  "name": "Potomac Falls"
}
{
  "address": {
    "city": "Frederick",
    "street": "9871 River Street",
    "zip": "221704"
  },
  "coordinate": {
    "latitude": 39.23576,
    "longitude": -77.4545
  },
  "id": 234,
  "name": "Monocacy River"
}

You can view the JSON after creating an item by selecting the item and then selecting the text view in the popup.

Note that the preceding JSON document is generic JSON. The actual JSON, as stored by DynamoDB (including datatypes), is as follows. Where the M, S, N, SS, etc. represent the element datatypes.

{
  "address": {
    "M": {
      "city": {
        "S": "Frederick"
      },
      "street": {
        "S": "9871 River Street"
      },
      "zip": {
        "S": "221704"
      }
    }
  },
  "coordinate": {
    "M": {
      "latitude": {
        "N": "39.23576"
      },
      "longitude": {
        "N": "-77.4545"
      }
    }
  },
  "id": {
    "N": "234"
  },
  "name": {
    "S": "Monocacy River"
  }
}

The DynamoDB datatypes are:

  • String = S,
  • Numbers = N,
  • Boolean = BOOL,
  • Binary = B,
  • Date = S (stored as a string),
  • String Set = SS,
  • Number Set = NS,
  • Binary Set = BS,
  • Map = M,
  • and List = L.

For example, in the following JSON document an observation’s address and coordinate are both Map datatypes, the city, street, zip are String datatypes, and the latitude and longitude are Number datatypes.

You can toggle between JSON and DynamoDB JSON in the popup window, as the following illustrates (note the DynamoDB JSON checkbox).

Create Observation Table

After creating the Station table we need to create the Observation table.

  • Create a new table named Observation.
  • Assign it a partition key of id and a sort key of stationid.

Composite Key (Partition Key & Sort Key)

The partition key is a table’s primary key and consists of a single attribute. DynamoDB uses this key to create a hash that determines the item’s storage. When used alone, the partition key uniquely identifies an item, as no two items can have the same partition key. However, when also defining a sort key, one or more items can have the same partition key, provided the partition key combined with the sort key is unique. Think of it as a compound key.

The Sort key helps DynamoDB more effectively store items, as it groups items with the same sort key together (hence the name sort key, as it sorts the items using this key).

An observation should have an id that identifies it and observations should be sorted by station, so we defined a stationid as the table’s sort key.

Create Sample Observations

As with the Station table, create some Observation items rather than define a schema.

  • Find three images, of small size, to use for this project. If you wish, use the three sample images from this tutorial’s Git project.
  • Navigate to Code Beautify’s Convert Your Image to Base64 webpage and convert the three images to a Base64 string.
  • Create a JSON list of observations.
  • Or, if you wish, simply use the JSON sampleData.json file provided in this tutorial’s Git project.

The following is a JSON list of four observations. The image base64 string is truncated so it can be easily displayed here. You can obtain the original file, named observations.json, from this tutorial’s Git project.

{
  [
    {
      "stationid": 221,
      "date": "1992-03-12",
      "time": "091312",
      "image": "/9j/4AAQSkZJRgABAQAAYABg <snip> kf/9k="
    },
    {
      "stationid": 221,
      "date": "1999-09-22",
      "time": "071244",
      "image": "/9j/4AAQSkZJ <snip> D9KhoA//2Q=="
    },
    {
      "stationid": 234,
      "date": "2001-11-09",
      "time": "111322",
      "image": "/9j/4AAQSkZ <snip> WoGf/9k="
    },
    {
      "stationid": 234,
      "date": "2013-01-12",
      "time": "081232",
      "image": "/9j/4AAQS <snip> q5//2Q=="
    }
  ]
}

Base64 Encoding

Images are binary. However, all binary can be represented by a String provided it is encoded and decoded correctly. Base64 is an encoding scheme that is converts binary to a string. It’s useful because it allows embedding binary data, such as an image, in a textual file, such as a webpage or JSON document. DynamoDB uses Base64 format to encode binary data to strings when transporting data and decode strings to binary data when storing the data. Therefore, the image sent to the Rest endpoints we create should be base64 encoded.

Create IAM Application User

Before beginning the Spring Boot application, we need a user with programatic access to the AWS DynamoDB API. If you are unfamiliar with IAM, the following introductory video should prove helpful. Otherwise, let’s create a user.

  • Navigate to the IAM Console and click Add user.
  • Create a new user named DynamoDBUser.
  • Assign DynamoDBUser with Programmatic access.
  • Create a new group named dynamo_users with AmazonDynamoDBFullAccess.
  • Assign DynamoDBUser to the dynamo_users group.
  • If you created the user correctly, you should see the following Summary screen.
  • Save the credentials file, credentials.csv, to your local hard-drive.

Spring Boot Application

Now that we have created the two needed tables and created a user we can begin the sample application. We create a Rest API for stations to save, retrieve, update, and delete observations. Not much explanation is devoted to Spring Boot, so if you have never created a Spring Boot Rest application you might consider completing a tutorial or two on Spring Boot and Rest. The following are links to two tutorials; however, there are many more on the web.

Setup Project

  • Create a new Project, I used Eclipse and created a new Maven application.
  • Modify the Maven POM file to match the following POM.
<project xmlns="http://maven.apache.org/POM/4.0.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
  <modelVersion>4.0.0</modelVersion>
  <groupId>com.tutorial.aws</groupId>
  <artifactId>DynamoDbTutorial</artifactId>
  <version>0.0.1-SNAPSHOT</version>
  <parent>
    <groupId>org.springframework.boot</groupId>
      <artifactId>spring-boot-starter-parent</artifactId>
      <version>2.0.5.RELEASE</version>
  </parent>
  <properties>
    <java.version>1.8</java.version>
  </properties>
  <dependencyManagement>
    <dependencies>
      <dependency>
        <groupId>software.amazon.awssdk</groupId>
	<artifactId>bom</artifactId>
	<version>2.5.25</version>
	<type>pom</type>
	<scope>import</scope>
      </dependency>
    </dependencies>
  </dependencyManagement>
  <dependencies>
    <dependency>
      <groupId>org.springframework.boot</groupId>
      <artifactId>spring-boot-starter-web</artifactId>
    </dependency>
    <dependency>
      <groupId>org.springframework.boot</groupId>
      <artifactId>spring-boot-starter-test</artifactId>
      <scope>test</scope>
    </dependency>
    <dependency>
      <groupId>com.jayway.jsonpath</groupId>
      <artifactId>json-path</artifactId>
      <scope>test</scope>
    </dependency>
    <dependency>
      <artifactId>auth</artifactId>
      <groupId>software.amazon.awssdk</groupId>
    </dependency>
    <dependency>
      <artifactId>aws-core</artifactId>
      <groupId>software.amazon.awssdk</groupId>
    </dependency>
    <dependency>
      <groupId>software.amazon.awssdk</groupId>
      <artifactId>auth</artifactId>
    </dependency>
    <dependency>
      <artifactId>dynamodb</artifactId>
      <groupId>software.amazon.awssdk</groupId>
    </dependency>
  </dependencies>
  <build>
    <plugins>
      <plugin>
        <groupId>org.springframework.boot</groupId>
	<artifactId>spring-boot-maven-plugin</artifactId>
      </plugin>
      <plugin>
        <groupId>org.apache.maven.plugins</groupId>
        <artifactId>maven-jar-plugin</artifactId>
        <version>3.1.1</version>
        <executions>
          <execution>
	    <phase>package</phase>
	    <goals>
              <goal>jar</goal>
	    </goals>
	    <configuration>
 	      <classifier>client</classifier>
	      <includes>
	        <include>**/factory/*</include>
	      </includes>
	    </configuration>
          </execution>
        </executions>
      </plugin>
    </plugins>
  </build>
</project>

In the POM we define the AWS Bill of Materials (BOM) and the required AWS libraries. Note that when using a BOM it is unnecessary to specify the library versions, as the BOM manages versions. We also define the Spring Boot libraries required.

  • Create an application.properties file in the resources folder. Open credentials.csv and add the credentials to the file with the following property names.
  • NOTE: THIS USER WAS DELETED BEFORE PUBLISHING THIS TUTORIAL.
cloud.aws.credentials.accessKey=AK <snip> WP
cloud.aws.credentials.secretKey=yLJJ <snip> asUR
cloud.aws.region.static=us-east-1

Create Spring Boot Application Class

  • Create a new class named SiteMonitorApplication in the com.tutorial.aws.dynamodb.application package.
  • Annotate the class with @SpringBootApplication annotation.
  • Create the main method and have it launch the Spring Boot application.
package com.tutorial.aws.dynamodb.application;

import org.springframework.boot.SpringApplication;
import org.springframework.boot.autoconfigure.SpringBootApplication;
import org.springframework.context.annotation.ComponentScan;

@SpringBootApplication
@ComponentScan({ "com.tutorial.aws.dynamodb" })
public class SiteMonitorApplication {

    public static void main(String[] args) {
        SpringApplication.run(SiteMonitorApplication.class, args);
    }
}

Create Observation Data Object

  • Create a class named Observation in the com.tutorial.aws.dynamodb.model package.
  • Create variables with the same names and types as in the JSON data created above.
package com.tutorial.aws.dynamodb.model;

import com.fasterxml.jackson.core.JsonProcessingException;
import com.fasterxml.jackson.databind.ObjectMapper;
import java.util.List;

public class Observation {
	private long stationid;
	private String date;
	private String time;
	private String image;
        private List<String> tags;
	
	public long getStationid() {
		return stationid;
	}

	public void setStationid(long stationid) {
		this.stationid = stationid;
	}

	public String getDate() {
		return date;
	}

	public void setDate(String date) {
		this.date = date;
	}


	public String getTime() {
		return time;
	}


	public void setTime(String time) {
		this.time = time;
	}

	public String getImage() {
		return image;
	}

	public void setImage(String image) {
		this.image = image;
	}

	public void setTags(List<String> tags) {
		this.tags = tags;
	}
	
	public List<String> getTags() {
		return this.tags;
	}
	
	@Override
	public String toString() {
		try {
			ObjectMapper mapper = new ObjectMapper();
			return mapper.writeValueAsString(this);
		} catch (JsonProcessingException e) {
			e.printStackTrace();
			return null;
		}
	}
}

The Observation object’s attributes are the same as in the JSON Observation document. Notice in the toString method we used an ObjectMapper from the Jackson library. We did not include this library in our POM, as the spring-boot-starter-web library includes this library.

The ObjectMapper maps JSON to Objects and Objects to JSON. It is how Spring Rest accomplishes this task. In the toString method we are telling the ObjectMapper instance to write the Observation object as a JSON string. For more on the ObjectMapper, here is a tutorial that explains the class in more depth: Jackson ObjectMapper.

Create Rest Controller

The Rest Controller provides the external API visible to Stations to send data to our application. Through the API, client applications will transmit data to the DynamoDB database. Different stations can develop its own client application in any language that supports Rest. The only requirement is that the station’s data follows the expected JSON format.

  • Note: we are violating the MVC Design pattern by putting data access directly in the Controller. Suspend disbelieve and ignore this anti-pattern.

Let’s create a Rest Controller to define our application’s API.

  • Create a class named ObservationApiController in the com.tutorial.aws.dynamodb.api package and annotate it with the @RestController annotation.
  • Assign it a top-level path of /observations.
  • Create a Rest endpoint for uploading a new Observation. Assign it the /observation mapping and name the method createObservation.
  • Have the method take an Observation as the request’s body.
  • Have the method print the uploaded Observation to the command-line.
package com.tutorial.aws.dynamodb.api;

import org.springframework.web.bind.annotation.PostMapping;
import org.springframework.web.bind.annotation.RequestBody;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RestController;
import com.tutorial.aws.dynamodb.model.Observation;

@RestController
@RequestMapping(value = "/observations")
public class ObservationApiController {

	@PostMapping("/observation")
	public void createObservation(@RequestBody Observation 
          observation) {
	  System.out.println(observation.toString());
	}	
}
  • Compile the application using Maven and start the application.
  • After the application starts, we can test using Postman.

Test using Postman

Postman is a useful tool for testing JSON endpoints. If you have never used Postman, you might consider completing a couple tutorials first.

  • Create a new request named AddObservation that exercises the Rest endpoint.
http://localhost:8080/observations/observation
  • Place one of the observations from the previously created JSON document in the request’s Body. Assign the type as JSON (application/json).
JSON Request in Postman for saving Observation.
  • Click Send to send the request to the Spring Rest endpoint. If everything is correct, you should see the Observation as JSON printed to the command-line.
{"stationid":221,"date":"1992-03-12", "time":"091312","image":"/9j/4AAQSkZJRgAB <snip> Wxkf/9k=","tags":null}
  • Copy the image base64 string and navigate to the CodeBeautify website’s Convert Your Base64 to Image webpage. Paste the string in the provided textarea and click Generate Image. If the base64 string was sent correctly, you should see the same image you sent to the Rest endpoint.

Create DynamoDB Client

Now that we have the basic Spring Boot application in place, we can start building the actual API to DynamoDB. But before working with DynamoDB, we need to create a DynamoDBClient instance.

  • Create a class named ObservationService in the com.tutorial.aws.dynamodb.service package.
  • Add the spring @Service annotation so spring sees this class as a controller.
  • Add the key and secretKey parameters and use the @Value annotation to indicate they are parameters from the application’s application.properties file (Spring Framework documentation).
  • Create a @PostConstruct and @PreDestroy methods (or implement a Spring InitializingBean).
  • Create a member variable entitled dynamoDbClient of type DynamoDbClient.
  • Instantiate and load the credentials for dynamoDbClient in the initialize method.
  • Close the dynamoDbClient in the preDestroy method.
package com.tutorial.aws.dynamodb.service;

import javax.annotation.PostConstruct;
import javax.annotation.PreDestroy;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.stereotype.Service;
import com.tutorial.aws.dynamodb.model.Observation;
import software.amazon.awssdk.auth.credentials.AwsBasicCredentials;
import software.amazon.awssdk.auth.credentials.StaticCredentialsProvider;
import software.amazon.awssdk.regions.Region;
import software.amazon.awssdk.services.dynamodb.DynamoDbClient;


@Service
public class ObservationService {

  @Value("${cloud.aws.credentials.accessKey}")
  private String key;

  @Value("${cloud.aws.credentials.secretKey}")
  private String secretKey;

  private DynamoDbClient dynamoDbClient;

  @PostConstruct
  public void initialize() {
    AwsBasicCredentials awsCreds = AwsBasicCredentials.create(key, secretKey);
    DynamoDbClient client = DynamoDbClient.builder()
        .credentialsProvider(StaticCredentialsProvider.create(awsCreds))
        .region(Region.US_EAST_1).build();

    this.dynamoDbClient = client;
  }

  @PreDestroy
  public void preDestroy() {
    this.dynamoDbClient.close();
  }
}

DynamoDBClient

The DynamoDBClient provides access to the DynamoDB API. All interaction with DynamoDB is done through this class. It has methods for for reading, writing, updating, and other interactions with DynamoDB tables and Items. For more information, refer to the API documentation.

Write Observation

Let’s first write an Observation to DynamoDB. Alternatively, you could say we PUT an item to DynamoB, as we are making an HTTP Put request to DynamoDB. We do this using the DynamoDBClient putItem method combined with a PutItemRequest.

Modify Service Class

  • Create a method named writeObservation that takes an Observation as a parameter.
  • Create a HashMap that uses String as the key and AttributeValue as the value.
  • Put each of the Observation variables into the HashMap, being sure the keys are correctly named. The keys should have exactly the same name as the JSON.
  • When creating the AttributeValueBuilder for each variable, ensure the correct datatype method is used.
  • Build a new PutItemRequest and then have dynamoDbClient call its putItem method to write the observation to the Observation DynamoDB table.
package com.tutorial.aws.dynamodb.service;

import java.util.HashMap;
import javax.annotation.PostConstruct;
import javax.annotation.PreDestroy;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.stereotype.Service;
import com.tutorial.aws.dynamodb.model.Observation;
import software.amazon.awssdk.auth.credentials.AwsBasicCredentials;
import software.amazon.awssdk.auth.credentials.StaticCredentialsProvider;
import software.amazon.awssdk.core.SdkBytes;
import software.amazon.awssdk.regions.Region;
import software.amazon.awssdk.services.dynamodb.DynamoDbClient;
import software.amazon.awssdk.services.dynamodb.model.AttributeValue;
import software.amazon.awssdk.services.dynamodb.model.PutItemRequest;

<snip>

public void writeObservation(Observation observation) {
   HashMap<String, AttributeValue> observationMap = new HashMap<String,
       AttributeValue>();

    observationMap.put("id", AttributeValue.builder()
      .s(observation.getStationid() + observation.getDate() + 
      observation.getTime()).build());

    observationMap.put("stationid", AttributeValue.builder()
      .n(Long.toString(observation.getStationid())).build());

    observationMap.put("date", AttributeValue.builder()
      .s(observation.getDate()).build());

    observationMap.put("time", AttributeValue.builder()
      .s(observation.getTime()).build());

    observationMap.put("image", AttributeValue.builder()
      .b(SdkBytes.fromUtf8String(observation.getImage())).build());

    if (observation.getTags() != null) {
      observationMap.put("tags", AttributeValue.builder()
        .ss(observation.getTags()).build());
    }

    PutItemRequest request = PutItemRequest.builder()
      .tableName("Observation").item(observationMap).build();

    this.dynamoDbClient.putItem(request);
  }
}

AttributeValue

There are four different AttributeValue classes in the DynamoDB Java API. Here we use the one in the software.amazon.awssdk.services.dynamodb.model package (api documentation). Remember, tables store items. An item is comprised of one or more attributes. An AttributeValue holds the value for an attribute. AttributeValue has a builder (api documentation) used to build an AttributeValue instance. An attribute value can be a string, number, binary data, list, or collection. You use the appropriate method corresponding to the datatype to set the AttributeValue object’s value. For instance, for a String use s(String value), binary use b(SdkBytes b), and for a collection of strings use ss(Collection ss). For a complete list, refer to the API documentation.

AttributeValue instances are placed in a Map, where the key is the attribute’s name in the database table. The Observation’s attributes are mapped using the appropriate builder methods.

  • Observation.id is a String, so we use,
.s(observation.getStationid() + observation.getDate() + 
       observation.getTime()).build()
  • The image, although encoded, is binary, and so we use,
observationMap.put("image", AttributeValue.builder()
      .b(SdkBytes.fromUtf8String(observation.getImage())).build());
  • The tags are an optional list of strings, so we wrap it in a conditional and use,
if (observation.getTags() != null) {
  observationMap.put("tags", AttributeValue.builder()
     .ss(observation.getTags()).build());
}

PutItemRequest

The PutItemRequest wraps the JSON request sent to the DynamoDBClient putItem method. A PutItemRequestBuilder builds a PutItemRequest. Above, we first added the table name, followed by the item to put. The item is a key-value map of the observation’s attributes. After building the PutItemRequest instance, the DynamoDBClient instance uses the request to write the observation to the DynamoDB Observation table.

PutItemRequest request = PutItemRequest.builder().tableName("Observation")
   .item(observationMap).build();
this.dynamoDbClient.putItem(request);

For more information, refer to the API documentation.

Create Rest Endpoint

  • Auto-wire the ObservationService to ObservationApiController using the @Autowired annotation.
  • Modify the saveObservation method to call the service’s writeObservation method.
package com.tutorial.aws.dynamodb.api;

import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.web.bind.annotation.PostMapping;
import org.springframework.web.bind.annotation.RequestBody;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RestController;

import com.tutorial.aws.dynamodb.model.Observation;
import com.tutorial.aws.dynamodb.service.ObservationService;

@RestController
@RequestMapping(value = "/observations")
public class ObservationApiController {
  @Autowired
  ObservationService observationService;
	
  @PostMapping("/observation")
  public void saveObservation(@RequestBody Observation observation) {
    this.observationService.writeObservation(observation);
  }	
}
  • Build the application and start the Spring Boot application.

Test with Postman

  • Return to the request created previously in Postman and click the Send button.
  • Go to the table in the AWS Console and click Items.
  • If everything was done correctly, you should see two items in the Observation table.
{
  "date": {
    "S": "1999-09-22"
  },
  "id": {
    "S": "2211999-09-22071244"
  },
  "image": {
    "B": "LzlqLzRBQ <snip> JRPT0="
  },
  "stationid": {
    "N": "221"
  },
  "tags": {
    "SS": [
      "observation",
      "river",
      "sample"
    ]
  },
  "time": {
    "S": "071244"
  }
}

Read Observation

Now that we have written an Observation to DynamoDB, let’s create a Rest endpoint that fetches an Observation.

Modify Service Class

  • Modify ObservationService to have a getObservation method that takes an observation’s id as a parameter.
  • Create a HashMap to store the AttributeValue collection. These are used to query the database.
  • Add the observationId to the HashMap.
  • Build a GetItemRequest, assigning the HashMap as the key.
  • Get the item from the response and use its values to create a new Observation.
  • As tags is a list of items, you must loop through them to add them to the Observation object instance.
public Observation getObservation(String observationId) {
  HashMap<String,AttributeValue> key = new 
    HashMap<String,AttributeValue>();

  key.put("id", AttributeValue.builder().s(observationId).build());
  GetItemRequest request = GetItemRequest.builder()
    .tableName("Observation").key(key).build();
  
  Map<String,AttributeValue> responseItem = 
  this.dynamoDbClient.getItem(request).item();

  Observation observation = new Observation();
  observation.setDate(responseItem.get("date").s());
  observation.setTime(responseItem.get("time").s());
  observation.setImage(responseItem.get("image").b().asUtf8String());
  observation.setStationid(Long.parseLong(responseItem
    .get("stationid").n()));
		
  if(responseItem.get("tags") != null && responseItem.get("tags")
    .ss().size() > 0) {

    HashSet<String> vals = new HashSet<>();
    responseItem.get("tags").ss().stream().forEach(x->vals.add(x));
    observation.setTags(vals);
  }
  return observation;
}

GetItemRequest

The GetItemRequest wraps a JSON Get request to DynamoDB. To fetch a particular Observation we must provide the id to the Get request. The key is a Map of AttributeValue items. In this case we added only one attribute, the id.

GetItemRequest request = GetItemRequest.builder()
    .tableName("Observation").key(key).build();

Create Rest Endpoint

  • Add a method named getObservation that takes the observation’s id as a path variable.
  • Call the ObservationService getObservation method and return the result.
@GetMapping("/observation/{observationid}")
public Observation getObservation(@PathVariable("observationid") String 
  observationId) {

  return this.observationService.getObservation(observationId);

}

Test With Postman

  • Create a new request in Postman with the following URL.
  • Add one of the ids from the observations added earlier.
http://localhost:8080/observations/observation/2211992-03-12091312
  • After creating the request, click Send and the following should appear as the response (if you used 211992-03-12091312) .
{
    "stationid": 221,
    "date": "1992-03-12",
    "time": "091312",
    "image": "/9j/4AA <snip> kf/9k=",
    "tags": [
        "rapids",
        "rocks",
        "observation",
        "cold"
    ]
}

Delete Observation

So far we have added and fetched an Observation to DynamoDB. Now let’s delete an Observation.

Modify Service Class

  • Add a deleteObservation method that takes an observation’s id as a parameter.
  • Create a HashMap to hold the attributes.
  • Build a new DeleteItemRequest and use the HashMap as the key.
  • Use the dynamoDbClient to delete the observation.
public void deleteObservation(String observationId) {
  HashMap<String,AttributeValue> key = new HashMap<>();
  key.put("id", AttributeValue.builder().s(observationId).build());
  DeleteItemRequest deleteRequest = DeleteItemRequest.builder()
    .key(key).tableName("Observation").build();
  this.dynamoDbClient.deleteItem(deleteRequest);
}

DeleteItemRequest

The DeleteItemRequest wraps a JSON Delete HTTP request. As with all requests, we use a builder. The builder uses the table and the key to delete the Observation.

Create Rest Endpoint

  • Create a new Rest endpoint to delete observations.
  • Have the observation’s id passed to the endpoint as a path variable only add /delete after the variable.
  • Call the ObservationService deleteObservation method.
@DeleteMapping("/observation/{observationid}/delete")
public void deleteObservation(@PathVariable("observationid") String 
  observationId) {
  this.observationService.deleteObservation(observationId);
}

Test with Postman

  • Create a new Request using Postman.
  • Assign it DELETE from the dropdown to indicate it is an Http Delete request.
http://localhost:8080/observations/observation/2211992-03-12091312/delete
  • Click Send and the record should be deleted. Navigate to the Items in the AWS Console to ensure the Observation was deleted.

Update Observation

An Observation can have one or more tags. This is something that seems likely to be added at a later date and/or modified. Let’s create an endpoint that allows adding/modifying an observation’s tags.

Update Service Class

  • Create a method named updateObservationTags that takes a list of tags and an observation id as parameters.
  • Create a HashMap to hold AttributeValue objects.
  • Use the AttributeBuilderValue builder to add the tags to the HashMap with :tagval as the key.
  • Create a second HashMap to hold the observation’s id.
  • Build an UpdateItemRequest that uses an update expression.
public void updateObservationTags(List<String> tags, String observationId) {
  HashMap<String, AttributeValue> tagMap = new HashMap<String, 
    AttributeValue>();

  tagMap.put(":tagval", AttributeValue.builder().ss(tags).build());
  HashMap<String, AttributeValue> key = new HashMap<>();
  key.put("id", AttributeValue.builder().s(observationId).build());

  UpdateItemRequest request = UpdateItemRequest.builder()
    .tableName("Observation").key(key)
    .updateExpression("SET tags = :tagval")
    .expressionAttributeValues(tagMap).build();

  this.dynamoDbClient.updateItem(request);
}

UpdateItemRequest

The DynamoDBClient instance uses the UpdateItemRequest to build the request to update the item. As with fetching and deleting, it needs a key to properly select the correct item. But it also needs the values to update. You provide an update expression and then provide the attributes. Note that the key for the attribute, :tagval, matches the expression. The request then uses the key and the update expression to update the item.

Add Rest Endpoint

  • Add an endpoint that takes the observation id as a path variable and a JSON array of tags as the request body.
  • Call the ObservationService updateObservationTags method.
@PostMapping("/observation/{observationid}/updatetags")
public void updateObservationTags(@PathVariable("observationid") String 
  observationId, @RequestBody List<String> tags) {
  
  this.observationService.updateObservationTags(tags, observationId);

}

Test With Postman

  • Create a new Http Post Request with the following URL.
  • Add the id value as a path parameter directly in the URL.
http://localhost:8080/observations/observation/2211992-03-12091312/updatetags
  • Add the following values as the Body and assign the type as JSON (application/json).
["observation","rocks","rapids","cold"]
  • Click Send and then navigate to the AWS Console to view observations. The Observation should have the tags added.

Batch Write Observations

Sometimes multiple items must be written to a database.

Update Service Class

public void batchWriteObservations(List<Observation> observations) {
  ArrayList<WriteRequest> requests = new ArrayList<>();
  HashMap<String, AttributeValue> observationMap = new HashMap<>();
  for(Observation observation : observations) {
    observationMap.put("id", AttributeValue.builder()
      .s(observation.getStationid() + observation.getDate() + 
      observation.getTime()).build());

    observationMap.put("stationid", AttributeValue.builder()
      .n(Long.toString(observation.getStationid())).build());

    observationMap.put("date", AttributeValue.builder()
      .s(observation.getDate()).build());

    observationMap.put("time", AttributeValue.builder()
      .s(observation.getTime()).build());

    observationMap.put("image", AttributeValue.builder()
      .b(SdkBytes.fromUtf8String(observation.getImage())).build());

    if (observation.getTags() != null) {
      observationMap.put("tags", AttributeValue.builder()
        .ss(observation.getTags()).build());
    }

    WriteRequest writeRequest = WriteRequest.builder()
      .putRequest(PutRequest.builder().item(observationMap)
      .build()).build();

    requests.add(writeRequest);
  }

  HashMap<String,List<WriteRequest>> batchRequests = new HashMap<>();
  batchRequests.put("Observation", requests);
  BatchWriteItemRequest request = BatchWriteItemRequest.builder()
    .requestItems(batchRequests).build();

  this.dynamoDbClient.batchWriteItem(request);
}

The DynamoDbClient batchWriteItem method takes a BatchWriteItemRequest as a parameter. The BatchWriteItem can write or delete up to 25 items at once and is limited to 16 MB of data. Note that it still makes as many calls as you have items; however, it makes these calls in parallel.

You create a List to hold the WriteRequest for each Observation. Each Observation is written to a Map as key-value pairs. The map is added to a WriteRequest, which is then added to the list until all observations are prepared as WriteRequest instances.

 WriteRequest writeRequest = WriteRequest.builder()
      .putRequest(PutRequest.builder().item(observationMap)
      .build()).build();

Each list of WriteRequest instances is added to another map. The table name is the key and the list is the values. In this way a single batch write could write to multiple tables. After creating the map of the lists of WriteRequest instances, the whole thing is used to create a BatchWriteItemRequest which is used by the DynamoDbClient batchWriteItem method.

 HashMap<String,List<WriteRequest>> batchRequests = new HashMap<>();
  batchRequests.put("Observation", requests);
  BatchWriteItemRequest request = BatchWriteItemRequest.builder()
    .requestItems(batchRequests).build();

Create Rest Endpoint

  • Add a Rest endpoint that takes a list of observations as JSON in the request body.
  • Call the batchWriteObservations method in the ObservationService.
@PostMapping("/observation/batch")
public void batchSaveObservation(@RequestBody List<Observation> 
  observations) {

  this.observationService.batchWriteObservations(observations);

}
  • Build and run the application.

Test With Postman

  • Create a new Post Request in Postman with the following URL.
http://localhost:8080/observations/observation/batch
  • Add the following to the Body and assign it the type JSON (application/json).
[
    {
      "stationid": 221,
      "date": "2007-12-12",
      "time": "180000",
      "image": "/9j/4AAQSkZJRgABAQA <snip> kf/9k="
    },
    {
      "stationid": 221,
      "date": "2009-05-22",
      "time": "043455",
      "image": "/9j/4AAQSkZJRgABAQAA <snip> /8AD9KhoA//2Q=="
    },
    {
      "stationid": 234,
      "date": "2019-10-18",
      "time": "121459",
      "image": "/9j/4AAQSkZJRgABAQA <snip> VWoGf/9k="
    },
    {
      "stationid": 234,
      "date": "2017-09-22",
      "time": "093811",
      "image": "/9j/4AAQSkZJRgAB <snip> 5//2Q=="
    }
  ]
  • Click Send then navigate to the AWS Console Observation table’s items and the observations should be added.

Conditionally Fetch Observations

A common requirement is to fetch records based upon certain criteria. For example, suppose we wish to fetch all observations belonging to a particular station. When using DynamoDB any variable used for a query must be indexed. So before creating a query, we first create an index on the Observation table’s stationid variable.

Create Index

  • Navigate to the Observation table in the AWS Console.
  • Click Create Index.
  • Select stationid as the index’s partition key and be certain to define it as a Number.
  • Click Create Index to create the index.

Secondary Indexes

Secondary Indexes allow retrieving data from a table using an attribute other than the primary key. You retrieve data from the index rather than the table. For more on DynamoDB secondary indexes, refer to the following article by LinuxAcademy: A Quick Guide to DynamoDB Secondary Indexes.

Update Service Class

public List<Observation> getObservationsForStation(String stationId){
  ArrayList<Observation> observations = new ArrayList<>();
  Condition condition = Condition.builder()
    .comparisonOperator(ComparisonOperator.EQ)
    .attributeValueList(AttributeValue.builder()
    .n(stationId).build()).build();

  Map<String, Condition> conditions = new HashMap<String, Condition>();
  conditions.put("stationid",condition);

  QueryRequest request = QueryRequest.builder().tableName("Observation")
    .indexName("stationid-index").keyConditions(conditions).build();

  List<Map<String, AttributeValue>> results = this.dynamoDbClient
    .query(request).items();
		
  for(Map<String,AttributeValue> responseItem: results) {
    Observation observation = new Observation();
    observation.setDate(responseItem.get("date").s());
    observation.setTime(responseItem.get("time").s());
    observation.setImage(responseItem.get("image").b().asUtf8String());
    observation.setStationid(Long.parseLong(
      responseItem.get("stationid").n()));
			
    if(responseItem.get("tags") != null && responseItem.get("tags").ss()
      .size() > 0) {
      HashSet<String> vals = new HashSet<>();
      responseItem.get("tags").ss().stream().forEach(x->vals.add(x));
      observation.setTags(vals);
    }
    observations.add(observation);
  }		
  return observations;		
}

First we created a Condition using its associated builder. The condition is “=<the station id passed to function>”.

Condition condition = Condition.builder()
    .comparisonOperator(ComparisonOperator.EQ)
    .attributeValueList(AttributeValue.builder()
    .n(stationId).build()).build();

We then added the Condition to a map and specified stationid as the key and condition as the value. We then built the QueryRequest using its associated builder.

 QueryRequest request = QueryRequest.builder().tableName("Observation")
    .indexName("stationid-index").keyConditions(conditions).build();

Create Rest Endpoint

  • Create a new Rest endpoint that specifies stationid as a path variable.
  • Call the ObservationService getObservationsForStation method.
@GetMapping("/station/{stationid}")
public List<Observation> getObservations(@PathVariable("stationid") String 
  stationId) {
  return this.observationService.getObservationsForStation(stationId);
}
  • Build and run the application.

Test With Postman

  • Create a new Get request in Postman that includes a station’s id in the url.
  • Click Send and the Observation items for station 221 should appear as the response body.
http://localhost:8080/observations/station/221
[
    {
        "stationid": 221,
        "date": "2009-05-22",
        "time": "043455",
        "image": "/9j/4AA <snip> 0/8AD9KhoA//2Q==",
        "tags": null
    },
    {
        "stationid": 221,
        "date": "2007-12-12",
        "time": "180000",
        "image": "/9j/4 <snip> /rn+q07/sHxfyNUK0Wxkf/9k=",
        "tags": null
    },
    {
        "stationid": 221,
        "date": "1992-03-12",
        "time": "091312",
        "image": "/9j/4AAQSkZJRgABAQAAYAB <snip> K0Wxkf/9k=",
        "tags": [
            "rapids",
            "rocks",
            "observation",
            "cold"
        ]
    },
    {
        "stationid": 221,
        "date": "1999-09-22",
        "time": "071244",
        "image": "/9j/4n0g27Qu <snip> A//2Q==",
        "tags": [
            "observation",
            "river",
            "sample"
        ]
    }
]

Further Topics

There are several topics not explored in this tutorial. First, you can scan a database table. When you scan the table you return all the items in the table. Second, this tutorial did not discuss conditionally updating or deleting items. However, the principles are the same as conditionally querying a table for items. Also, it is helpful to explore the command-line examples for working with DynamoDB, as they help understand the SDK. Finally, we did not cover the Java 1.1 AWS SDK.

From Java 1.1 AWS SDK to Java 2 AWS SDK

There are many more examples and tutorial on the Web using the Java 1.1 API rather than the Java 2 API. However, the primary difference between the two versions is the builder pattern. Many, if not most, of the Java 1.1 tutorials remain useful. The pattern is the same:

  • create a request type
  • setup the request with the desired parameters,
  • pass the request to the DynamoDB client,
  • obtain the result.

In the Java 1.1 SDK you perform these steps using constructors and setters and getters. In the Java 2 SDK you use builders. Practically all classes in the Java 2 AWS SDK use builders. Use this as a starting point if you have a particularly good tutorial using the Java 1.1. SDK. Although not foolproof, doing this has helped me translate many Java 1.1. examples to Java 2 SDK.

Further Resources

  • More on DynamoDB design patterns.
  • This tutorial, although it uses the Java 1 AWS API, is a very good introduction covering the same topics in this tutorial. Just remember, think builder, although the techniques in the API are the same, the Java 2 version of the API uses builders extensively.

Conclusion

In this tutorial we explored the lower-level API of the Java 2 SDK by using the AWS DynamoDB Java API. We wrote an item, updated an item, deleted an item, and batch uploaded items. We also explored conditionally querying items.

As with all of the SDK, it is based upon builders, requests, and the client. You build a request to pass to the DynamoDBClient which in turn returns a response. You do not create a new instance of a request and set properties via setters, but rather, you use a builder to build a request.

DynamoDB is a non-relational database and so you cannot just write a conditional query on any field. You can only use fields that are indexed in a query. This seems logical if you consider that DynamoDB is designed for massive amounts of data that is relatively unstructured.

Git-hub Project