AWS Compute Blog

Application analytics pipeline with Amazon EventBridge

February 12, 2024: Amazon Kinesis Data Firehose has been renamed to Amazon Data Firehose. Read the AWS What’s New post to learn more.

This post is courtesy of Rajdeep Tarat, Solutions Architect and Venugopal Pai, Solutions Architect

Customers across industry verticals collect, analyze, and derive insights from end-user application analytics using solutions such as Google Analytics and MixPanel. While these solutions provide built-in dashboards for marketing analytics, it can be difficult to reuse the raw event data.

Setting up a pipeline to move the raw event data into AWS opens up possibilities for various rule-based, statistical, and machine learning algorithms to derive deep insights about end-user behavior. Additionally, the raw event data can be enriched with other transactional data points available within the customer’s AWS environment.

This post uses the Segment Partner integration in Amazon EventBridge to pipe the data into your AWS environment. Segment allows you to collect, unify, and connect end-user application analytics into AWS using Amazon EventBridge as a destination.

Segment already supports direct, optimized connections to many AWS services such as Amazon Redshift, Amazon Personalize, Amazon Kinesis, Amazon Kinesis Data Firehose, AWS Lambda, and Amazon S3. The EventBridge destination is a good choice for customers who want the flexibility and centralization that EventBridge offers.

EventBridge makes it easy to build scalable event-driven application pipelines by handling event ingestion, delivery, security, authorization, and error handling for you. The architecture of this pipeline is shown below:Segment architecture

In the diagram, end-user applications send the data into Segment, which is routed to each of the configured destinations (for example, EventBridge). Once the data reaches EventBridge, it is again routed to multiple targets. With this approach you can continue using existing solutions supported by Segment in parallel to the Amazon EventBridge destination.

This architecture makes the pipeline highly extensible and modular. Firstly, you can configure multiple Segment destinations to fan out the event data into other existing solutions in parallel to EventBridge. Marketing teams can continue to use their existing tools without any disruptions while the data is seamlessly pumped into AWS. Within the AWS Cloud, EventBridge can again route the event data to up to five targets per rule.

The following section provides a walkthrough of setting up the Segment integration with EventBridge, and configuring two targets within the AWS Cloud.

  • The first target uses an Amazon Kinesis Data Firehose to deliver the data to an S3 bucket. From the S3 bucket, multiple AWS services can use the data (learn more about using S3 as a data lake).
  • The second target posts the event data to an SNS topic. From here, the data can be consumed by subscribers for the topic.

Walkthrough

To set up the pipeline, you must configure the Segment partner integration in EventBridge, and then set up the targets where analytics data is sent.

Amazon EventBridge – Segment partner integration:

  1. From the Amazon EventBridge console, navigate to the Partner Event Sources > Segment setup page. Copy your AWS Account ID from here.
    Segment setup
  2. On the Segment destination setup page, use the Amazon EventBridge integration. Enter the AWS Account ID and select a Region (learn more about setting up a Segment destination).
    EventBridge settings

Create the event bus:

  1. After linking the Segment Destination with the AWS Account ID, send a test event from Segment to create a Partner Event Source. This also creates an Event bus with the same name. This is done by firing a test event from the Event Tester in the Segment Dashboard.
    Event Tester
  2. After the first test event is fired, the Partner Event Source and the corresponding event bus is created in the EventBridge console.
    Partner event sources

Create rules:

  1. A rule watches for incoming events and routes them to specific targets that are configured. Start by creating a new rule and entering a name.
    Creating rules
  2. For Event Pattern, select the Predefined pattern by Service, and select Service Partners > Segment.
    Define event pattern
  3. Under Select event bus, select the Custom or partner event bus, and the name of the event bus created.
    Select event bus

Configuring multiple targets for the event bus:

  1. For Kinesis streams, select Kinesis stream from the Target dropdown, and the name of the stream. For more details on creating a Kinesis data stream, read this documentation.
    Select targets
  2. For SNS topic, choose Add Target and repeat the same steps to add an SNS topic instead. For more details on creating an SNS topic, read this documentation.
    SNS as target
  3. You can optionally tag the resource, then choose Create.

The pipeline is ready to send data to the targets configured via the event bus. You can now send test events from the Segment dashboard and monitor Kinesis Data Firehose or by setting up subscribers for the SNS topic.

Conclusion

This post shows how customers can capture end-user application analytics using the partner solution Segment in real time, and ingest data into Amazon EventBridge. The data routing is made extensible using multiple Segment destinations (for third-party solutions), and using multiple rules in EventBridge (for multiple destinations within the AWS Cloud).

To learn more about Amazon EventBridge integrations, read the EventBridge documentation.