Event-driven applications

Developing applications on AWS

Ricardo Sueiras

Principal Technologist

What is an event?

 

• A message is a unit of work sent between components.
• An event records that something already happened.
• Services communicate by publishing events.
• Other services react independently to those events.
• Decoupling reduces dependencies and improves scalability.

Event producers

 

• Producers create and publish events when actions occur.
• Both AWS services and custom applications can produce events.
• Uploading a file to Amazon S3 is one example.
• Amazon S3 then emits an object-created event.

Event consumers

 

• Consumers receive and process events asynchronously.
• Event sources use different delivery models.
• Push-based: events are delivered immediately.
• Poll-based: consumers retrieve events when they poll.

Event source mapping

• AWS Lambda is a common event consumer.
• It polls sources and retrieves batches of records automatically.

 

Tuning Lambda event source mappings

• Tune event source mappings to improve throughput.
• Adjust batch size to control records per invocation.
• Configure retry controls for failed records.
• Use partial batch responses on supported sources.
• ReportBatchItemFailures avoids reprocessing successful records.

 

Amazon EventBridge

• EventBridge routes events between services and applications.
• It enables dynamic routing without tightly coupling services.
• Applications publish their events to EventBridge.
• Routing rules decide which targets receive each event.

 

EventBridge capabilities

 

• EventBridge offers several routing capabilities.
• Rules fan one event out to many targets.
• Scheduled rules remove the need for cron servers.
• Schedules support rate(), at(), and cron() expressions.
• Pipes provide point-to-point integration.
• Pipes also enable filtering and transformation.

Event buses

 

• Different event buses optimize how events are routed.
• The default bus receives events from AWS services.
• Custom buses carry application-specific events.
• Partner buses receive events from third-party providers.
• Rules sit between each bus and its targets.

Schemas discovery

 

• Producers and consumers often evolve independently.
• Understanding event structure is essential.
• Schema discovery captures the structure of incoming events.
• Use it to inspect event payloads.
• Use it to validate integrations.
• Use it to generate bindings for supported languages.

Schema registry

• The schema registry stores your event schemas.
• It helps developers understand event structure and generate bindings.

 

Event filtering

• Filtering reduces unnecessary processing.
• Define matches with JSON event patterns.
• Patterns match only the specific events you want.
• Downstream systems then process only relevant events.

 

Event ordering

• Event ordering is not always guaranteed.
• Some services may process events out of sequence.
• Use ordering controls to enforce sequence explicitly.

 

Archiving and replaying events

 

• EventBridge can archive and later replay events.
• Replay events to recover from failures.
• Replay events to test new consumers.
• Replay events to rebuild downstream systems.

Error handling

 

• Event-driven apps must handle errors safely.
• Failures include duplicate events.
  • Failed processing.
  • Unavailable consumers.
• Design consumers to be idempotent wherever possible.

DLQs and destinations

 

• Use DLQs to isolate repeatedly failing events.
• In EventBridge, DLQs are configured per rule target, not per function.
• Lambda Destinations extend this routing.
• Destinations route both success and failure results.

Testing and validating events

• Testing event-driven apps validates routing.
• It also validates event filtering.
• It also validates event retries.
• It confirms processing across distributed systems.
• Use AWS SAM to generate sample test events.
• Sample events replicate the exact JSON output.

 

What to test

• SAM lets you test locally before deployment.
• Validate event pattern matching.
• Validate retry behavior.
• Validate throttling scenarios.
• Validate duplicate event delivery.
• Validate idempotent consumer behavior.

 

Let's practice!

Developing applications on AWS