DEV Community

Cover image for ๐Ÿ’– AWS CDK 101 - ๐ŸŒ๏ธโ€โ™€๏ธ Scalable event-driven processing using Eventbridge and SQS
Aravind V
Aravind V

Posted on • Updated on • Originally published at devpost.hashnode.dev

๐Ÿ’– AWS CDK 101 - ๐ŸŒ๏ธโ€โ™€๏ธ Scalable event-driven processing using Eventbridge and SQS

๐Ÿ”ฐ Beginners new to AWS CDK, please do look at my previous articles one by one in this series.

If in case missed my previous article, do find it with the below links.

๐Ÿ” Original previous post at ๐Ÿ”— Dev Post

๐Ÿ” Reposted previous post at ๐Ÿ”— dev to @aravindvcyber

In this article, let us refactor our existing counter construct which is triggering the current synchronous backend processor into an event-driven scalable solution by introducing an event bridge that pushes new messages into the dedicated queue which can be processed by a backend processor which writes the data to dynamodb.

Benefits achieved in this approach ๐Ÿšฃโ€โ™€๏ธ

  • This will help us achieve a layer of decoupling between counter and backend processors.
  • By not invoking any new compute-intensive synchronous tasks like vertically adding a new lambda backend, so that our counter lambda can complete soon.
  • Processing time of the counter lambda, by not waiting for the time our synchronous backend takes for its processing before the parent task is complete.
  • Avoid scenarios like, we may end up when running 1000 lambda concurrently running by half, as we remove another lambda dependency and we could optimize our reservation on the lambda concurrency.
  • We could also set a predefined number of batched lambda processors for the final backend (here 2 is our batch limit) and reuse our unreserved lambda concurrency limit in our other solutions efficiently.
  • Continuous batch processing, keeps our lambda warm most of the time, whereas irregular spikes in lambda concurrency, may at times introduce a cold start more often.

New construct ๐Ÿšง

We will start by creating a new file under the constructs folder like constructs/event-counter-bus.ts.

So, why do we create a construct, that's because we could reuse this elsewhere in various other cases. Also when I demonstrate a new construct, it will be more generalized and easy to follow with even the limited idea about the actual functional stack, which we have built up in our previous articles.

Counter with Bus construct ๐Ÿšดโ€โ™‚๏ธ

Also, you can simply duplicate our previous construct constructs/event-counter.ts and start overwriting it as per our new requirement.

As like every other construct which we have created previously, let us start by importing the necessary libraries and interfaces for our props as shown below.

import { IEventBus } from 'aws-cdk-lib/aws-events';


export interface BusProperties {
  bus: IEventBus,
  props: {
    DetailType: string,
    EventBusName: string,
    Source: string,
  }
}

export interface EventCounterProps {
  /** the Event bus which we will use to send our messages to queue**/
  downstreamBus: BusProperties,
  //** refer our previous construct
  ////////////////////////
}
Enter fullscreen mode Exit fullscreen mode

Counter-Bus handler specification ๐Ÿšตโ€โ™‚๏ธ

We need to change our handler to event-counter-bus.counter and add some environment variables as shown below.

  • DetailType: downstreamBus.props.DetailType,
  • EventBusName: downstreamBus.props.EventBusName,
  • Source: downstreamBus.props.Source,
const eventCounterFn = new lambda.Function(this, 'EventCounterHandler', {
        runtime: lambda.Runtime.NODEJS_14_X,
        handler: 'event-counter-bus.counter',
        code: lambda.Code.fromAsset('lambda'),
        environment: {
          DetailType: downstreamBus.props.DetailType,
          EventBusName: downstreamBus.props.EventBusName,
          Source: downstreamBus.props.Source,
          EVENT_COUNTER_TABLE_NAME: Counters.tableName
    },
    logRetention: logs.RetentionDays.ONE_MONTH,
});
Enter fullscreen mode Exit fullscreen mode

Counter-Bus supporting lambda ๐Ÿšดโ€โ™€๏ธ

Now let us write our new lambda function
in the folder lambda as follows:
event-counter-bus.ts

Here as well we will be duplicating event-counter.ts and overwriting it a bit to get our desired effect.

Here will be only replacing the invoke lambda section with the below code to send an event to our event bus which we will create shortly inside our stack.


////// and the rest above in the previous file
  let resp = { Payload: ""};
  let output = { FailedEntryCount: 0}
  let data: PutEventsResultEntry[] | undefined = undefined;

  try {
    const msg: string = JSON.stringify({message});

    const eventData: EventBridge.PutEventsRequest = {
      Entries: [
        {
          Detail: msg,
          DetailType: process.env.DetailType || '',
          EventBusName: process.env.EventBusName || '',
          Source: process.env.Source || '',
        },
      ],
    };
   var proc: PromiseResult<PutEventsResponse, AWSError> = await eventBridge.putEvents(eventData).promise();
   output.FailedEntryCount = Number(proc.FailedEntryCount?.toString());
   resp.Payload = JSON.stringify(proc.Entries)

   data = proc.Entries

  } catch (err) {
    console.log(JSON.stringify(err));
    resp.Payload = JSON.stringify(err);
    ////// and the rest
Enter fullscreen mode Exit fullscreen mode

Before this code block, we to have import the below modules to facilitate this and removed the unwanted modules in this lambda so that we have better memory usage.

import { PutEventsResponse, PutEventsResultEntry } from "@aws-sdk/client-eventbridge";
import { AWSError, EventBridge } from "aws-sdk";
import { PromiseResult } from "aws-sdk/lib/request";

const eventBridge = new EventBridge({ region: "ap-south-1" });
Enter fullscreen mode Exit fullscreen mode

And the final return statement updated as shown below.

return {
      statusCode: 200,
      headers: { "Content-Type": "text/json" },
      body: JSON.stringify({data})
};
Enter fullscreen mode Exit fullscreen mode

Asynchronous Processor lambda ๐Ÿฆ™

Now let us create a simple processor lambda which will be invoked in a batch of 2 from the queue which we will define shortly.

In our case, we are reading the message from the queue and then we are extracting some information and writing this into dynamodb and thus we complete the simple processing step, which is more or less straightforward.

import { PutItemInput } from "aws-sdk/clients/dynamodb";

const { DynamoDB } = require("aws-sdk");

exports.processor = async function (event: any) {
  const dynamo = new DynamoDB();
  let result: any | undefined = undefined;
  await Promise.all(event.Records.map(async (msg: any) => {
    console.log("Received message:", JSON.stringify(msg, undefined, 2));
    const content = JSON.parse(msg.body);
    const putData: PutItemInput = {
      TableName: process.env.MESSAGES_TABLE_NAME || "",
      Item: {
        messageId: { S: content.id },
        createdAt: { N: msg.attributes.ApproximateFirstReceiveTimestamp },
        event: { S: msg.body },
      },
      ReturnConsumedCapacity: "TOTAL",
    };
    try {
      result = await dynamo.putItem(putData).promise()
    } catch (err) {
      console.log(err)
    }    
  }));

  return {
    statusCode: 200,
    headers: { "Content-Type": "text/json" },
    body: {
      ProcessorResult: `Message Processed : ${JSON.stringify({ result })}\n`,
    },
  };
};

Enter fullscreen mode Exit fullscreen mode

EventBridge setup - Common Event Bus ๐Ÿป

Now let us edit our common-event-stack.ts which we have created earlier in this series.

Let us define a new event bridge as follow.

const commonBus = new EventBus(this, "CommonEventBus", {
      eventBusName: "CommonEventBus",
});

commonBus.applyRemovalPolicy(RemovalPolicy.DESTROY);

Enter fullscreen mode Exit fullscreen mode

event bridge bus

Let us import the new construct which we have defined as follows with a bunch of other modules which we may require to define and configure the event bridge.


import { EventBus, Rule } from "aws-cdk-lib/aws-events";
import * as eventTargets from "aws-cdk-lib/aws-events-targets";
import { DeadLetterQueue, Queue } from "aws-cdk-lib/aws-sqs";
import { SqsEventSource } from "aws-cdk-lib/aws-lambda-event-sources";
import * as dynamodb from "aws-cdk-lib/aws-dynamodb";
import { EventCounterBus } from "../constructs/event-counter-bus";
Enter fullscreen mode Exit fullscreen mode

Initialise the construct ๐Ÿฐ

Now let us initialize the event-counter-bus construct and most importantly don't forget to grant privileges for PutEvents to the event bus created above.

const eventCounterBus = new EventCounterBus(this, "eventEntryCounterBus", {
      downstreamBus: {
        bus: commonBus,
        props: {
          DetailType: "CommonEvent",
          EventBusName: "CommonEventBus",
          Source: "com.devpost.commonevent",
        },
      },
      tableName: "Event Counters",
      partitionKeyName: "Counter Name",
    });
commonBus.grantPutEventsTo(eventCounterBus.handler);
Enter fullscreen mode Exit fullscreen mode

Queuing ๐Ÿชถ

A Queue helps to decouple and scale microservices, distributed systems, and serverless applications. SQS eliminates the complexity and overhead associated with managing and operating message-oriented middleware, and empowers developers to focus on differentiating work.

  • Standard queues offer maximum throughput, best-effort ordering, and at-least-once delivery.
  • SQS FIFO queues are designed to guarantee that messages are processed exactly once, in the exact order that they are sent.

Dead letter Queue ๐Ÿฆฉ

Before we define the various queue, we will be using in this article let us understand a DLQ, which can come very useful when we have some issues with our solution. The dead queue is supposed to grab these errored or failed messages, which could not be processed further in our pipeline. This can be later inspected to identify the issue. Normally creating this before creating an actual queue, makes good sense, as we did in this article.

Event Target DLQ ๐Ÿฆ‰

This will be used to buffer messages while we encounter issues in filtering via the event rule.

const commonEventTargetDLQ: DeadLetterQueue = {
      queue: new Queue(this, "commonEventTarget-DLQ", {
        retentionPeriod: Duration.days(14),
        removalPolicy: RemovalPolicy.DESTROY,
        queueName: "commonEventTarget-DLQ",
      }),
      maxReceiveCount: 100,
};
Enter fullscreen mode Exit fullscreen mode

Event Processor DLQ ๐Ÿฆข

DLQ will be used to store failed messages, while we encounter issues when the processor fails to successfully process the message from the standard queue after a certain retry limit is met. This is sort of a secondary storage area for quick inspection

const commonEventProcessorQueueDLQ: DeadLetterQueue = {
      queue: new Queue(this, "commonEventProcessorQueueDLQ", {
        retentionPeriod: Duration.days(14),
        removalPolicy: RemovalPolicy.DESTROY,
        queueName: "commonEventProcessorQueueDLQ",
      }),
      maxReceiveCount: 100,
};
Enter fullscreen mode Exit fullscreen mode

Processor DLQ

Queue for our lambda processor ๐Ÿฆ†

This queue will be used to poll from and process the messages in a batch by the processor lambda as follows.

const commonEventProcessorQueue = new Queue(
      this,
      "commonEventProcessorQueue",
      {
        retentionPeriod: Duration.days(5),
        removalPolicy: RemovalPolicy.DESTROY,
        deliveryDelay: Duration.seconds(3),
        queueName: "commonEventProcessorQueue",
        visibilityTimeout: Duration.minutes(1),
        deadLetterQueue: commonEventProcessorQueueDLQ,
      }
);
Enter fullscreen mode Exit fullscreen mode

Queues and DLQ

Event Rule Target ๐Ÿ˜

We will define a new Target for our rules, we will define later in this article. This queue is a standard buffer area where our messages can reside for a period with an optional dead letter queue defined above commonEventProcessorQueueDLQ

const commonEventQueueTarget = new eventTargets.SqsQueue(
      commonEventProcessorQueue,
      {
        retryAttempts: 3,
        deadLetterQueue: commonEventProcessorQueueDLQ.queue,
      }
);
Enter fullscreen mode Exit fullscreen mode

Event Bridge Rule ๐Ÿค

Now it is time to create an event-bridge rule which will be used for filtering the specific messages based on eventPattern we specify, here in this case we only use source but there are lots more including the DetailType, by which we could leverage to push messages to the specific target.

We have chosen the Queue which we defined as above us our target commonEventQueueTarget, but there are more connections which we do here, including invoking a new lambda, but to make it more scalable we brought in a queue which will be further vertically integrated with a lambda.

event bridge rule targets

const eventRule = new Rule(this, `CommonEventProcessorRule`, {
      eventBus: commonBus,
      eventPattern: { source: [`com.devpost.commonevent`] },
      //targets: [commonEventTarget],
      targets: [commonEventQueueTarget],
      ruleName: "CommonEventProcessorRule",
});

eventRule.applyRemovalPolicy(RemovalPolicy.DESTROY);
Enter fullscreen mode Exit fullscreen mode

event bridge rule

Event bridge rule definition ๐Ÿฆซ

event bridge rule def

initializing Processor lambda ๐Ÿปโ€โ„๏ธ

Here, we could also add the event-processor.ts lambda which we have scripted earlier in this article.


const eventProcessor = new lambda.Function(this, "EventProcessorHandler", {
      runtime: lambda.Runtime.NODEJS_14_X,
      code: lambda.Code.fromAsset("lambda"),
      handler: "event-processor.processor",
      logRetention: logs.RetentionDays.ONE_MONTH,
      environment: {
        MESSAGES_TABLE_NAME: envParams.messages.tableName || "",
      },
});

eventProcessor.applyRemovalPolicy(RemovalPolicy.DESTROY);
Enter fullscreen mode Exit fullscreen mode

Queue lambda

Linking Processor Queue to Processor lambda ๐Ÿฆ

Now link the new queue created to the above lambda processor by using the addEventSource function as follows.

eventProcessor.addEventSource(
      new SqsEventSource(commonEventProcessorQueue, {
        batchSize: 2,
})
);
Enter fullscreen mode Exit fullscreen mode

New Dynamodb table ๐Ÿฆค

Now let us add a new dynamodb table which we have earlier mentioned inside the event-processor.ts file.

const messages = new dynamodb.Table(this, "MessagesTable", {
      tableName: process.env.messagesTable,
      sortKey: { name: "createdAt", type: dynamodb.AttributeType.NUMBER },
      partitionKey: { name: "messageId", type: dynamodb.AttributeType.STRING },
      encryption: dynamodb.TableEncryption.AWS_MANAGED,
      readCapacity: 5,
      writeCapacity: 5,
});

messages.grantReadWriteData(eventProcessor);

Enter fullscreen mode Exit fullscreen mode

Dynamodb table view on aws console ๐Ÿธ

Dynomodb table

dynamodb description

Postman testing of the solution ๐Ÿฆƒ

Postman Event post

Lookup inside dynamodb using a query on the partitionkey ๐Ÿฆ‰

MessageId in dynamodb

Simple full table view with sort ๐Ÿฆ˜

Refer to our previous articles on using cdk-dynamodb-table-viewer

const tblViewer2 = new TableViewer(this, "Messages-", {
      title: "Messages from Dynamodb",
      table: messages,
      sortBy: "-createdAt"
});
Enter fullscreen mode Exit fullscreen mode

Messages table viewer

We will be adding more connections to our stack and making it more usable in the upcoming articles by creating new constructs, so do consider following and subscribing to my newsletter.

โญ We have our next article in serverless, do check out

https://dev.to/aravindvcyber/aws-cdk-101-statemachine-and-stepfunctions-replacing-our-sqs-based-lambda-trigger-3fg7

๐ŸŽ‰ Thanks for supporting! ๐Ÿ™

Would be great if you like to โ˜• Buy Me a Coffee, to help boost my efforts.

Buy Me a Coffee at ko-fi.com

๐Ÿ” Original post at ๐Ÿ”— Dev Post

๐Ÿ” Reposted at ๐Ÿ”— dev to @aravindvcyber


๐Ÿ’– AWS CDK 101 - ๐ŸŒ๏ธโ€โ™€๏ธ Scalable event-driven processing using Eventbridge and SQS#serverless #aws #dynamodb #awscdk #messagequeuehttps://t.co/cRb8hpHkxG

โ€” Aravind V (@Aravind_V7) April 11, 2022

Top comments (0)