Mastering Millisecond Latency with Event-Driven Architecture

Mastering Millisecond Latency and Millions of Events

The Amazon Key team successfully transformed their tightly coupled monolithic system into a resilient, event-driven solution using Amazon EventBridge, achieving significant improvements in reliability, scalability, and maintainability. By addressing the challenges of service coupling, loose event schemas, and inconsistent event routing, the team was able to process 2000 events/second with a 99.99% success rate and 80ms p90 latency.

Why This Matters

The Amazon Key team’s legacy architecture faced significant challenges stemming from its tightly coupled design, leading to system fragility and instability. The implementation of a single-bus, multi-account pattern and the development of a schema repository, client library, and subscriber constructs library enabled the team to achieve improved reliability, scalability, and maintainability. This technical reality highlights the importance of adopting well-designed event-driven architecture patterns to overcome the limitations of traditional monolithic systems.

Key Insights

• Amazon EventBridge provides a powerful foundation for event-driven architectures, enabling the capture of event structures in the schema registry and maintaining versions as events evolve over time: [Amazon Web Services, 2026]
• A custom schema repository with client-side validation is essential for large-scale event-driven architectures, providing immediate validation feedback and maintaining standards for schema versioning and release management: [Amazon Key Team, 2026]
• The use of a client library and subscriber constructs library simplifies and standardizes the integration process with the central event bus, enabling teams to focus on their core business logic: [AWS Cloud Development Kit, 2026]

Working Example

import boto3

# Create an EventBridge client
eventbridge = boto3.client('events')

# Define an event
event = {
    'Source': 'aws.key',
    'Resources': [],
    'DetailType': 'Key Event',
    'Detail': '{"key": "value"}'
}

# Put the event into the event bus
response = eventbridge.put_events(Entries=[event])

# Print the response
print(response)

Practical Applications

• Use Case: Amazon Key team uses event-driven architecture to process millions of events, achieving millisecond latency and high reliability.
• Pitfall: Failing to implement a well-designed event-driven architecture can lead to system fragility, instability, and poor scalability, resulting in significant costs and revenue losses.

References:

• https://aws.amazon.com/blogs/architecture/mastering-millisecond-latency-and-millions-of-events-the-event-driven-architecture-behind-the-amazon-key-suite/
• https://docs.aws.amazon.com/eventbridge/latest/userguide/eb-what-is.html
• https://docs.aws.amazon.com/cdk/latest/guide/getting_started.html