Event-Driven Serverless Architecture

By Michael Samuel Naeem · July 22, 2025 · 5 min read

The order service POSTed synchronously to inventory, email, and analytics—800ms p99 and cascading failures when analytics timed out. Event-driven serverless decouples producers from consumers: publish OrderPlaced, let Lambdas react independently, scale per handler, and retry without blocking checkout. The complexity moves to idempotency, ordering guarantees, and observability across invisible call chains.

Event envelope

{
  "event_id": "evt_01JXYZ",
  "type": "order.placed",
  "version": "1",
  "timestamp": "2025-07-22T10:00:00Z",
  "source": "orders-service",
  "data": {
    "order_id": "ord_991",
    "customer_id": "cust_42",
    "total_cents": 15000
  }
}

Schema registry (EventBridge Schema Registry, AsyncAPI) validates compatibility on publish.

Validate this in staging with production-like data volume before declaring done. Capture metrics baseline the week before change and compare for seven days after—subtle regressions hide in aggregates until a large tenant hits the path. Update the on-call runbook with the failure signature and rollback command so responders need not rediscover steps during an incident.

EventBridge rule example

EventPattern:
  source: ["orders-service"]
  detail-type: ["order.placed"]
Targets:
  - Arn: !GetAtt InventoryQueue.Arn
  - Arn: !GetAtt EmailQueue.Arn

Each target gets SQS buffer absorbing Lambda scale bursts.

Validate this in staging with production-like data volume before declaring done. Capture metrics baseline the week before change and compare for seven days after—subtle regressions hide in aggregates until a large tenant hits the path. Update the on-call runbook with the failure signature and rollback command so responders need not rediscover steps during an incident.

Lambda consumer with idempotency

def handler(event, context):
    for record in event["Records"]:
        body = json.loads(record["body"])
        event_id = body["event_id"]
        if already_processed(event_id):
            continue
        process_order(body["data"])
        mark_processed(event_id)

Use conditional PutItem on DynamoDB partition key event_id.

Validate this in staging with production-like data volume before declaring done. Capture metrics baseline the week before change and compare for seven days after—subtle regressions hide in aggregates until a large tenant hits the path. Update the on-call runbook with the failure signature and rollback command so responders need not rediscover steps during an incident.

Dead-letter queues

RedrivePolicy:
  deadLetterTargetArn: !GetAtt OrderDLQ.Arn
  maxReceiveCount: 5

Alert on DLQ depth. Replay tool reprocesses after fix with same idempotency guard.

Validate this in staging with production-like data volume before declaring done. Capture metrics baseline the week before change and compare for seven days after—subtle regressions hide in aggregates until a large tenant hits the path. Update the on-call runbook with the failure signature and rollback command so responders need not rediscover steps during an incident.

Ordering tradeoffs

Standard SQS preserves rough order but not strictly. FIFO queues serialize per MessageGroupId (e.g., order_id) with throughput limits. EventBridge does not guarantee order—design for out-of-order arrival.

Validate this in staging with production-like data volume before declaring done. Capture metrics baseline the week before change and compare for seven days after—subtle regressions hide in aggregates until a large tenant hits the path. Update the on-call runbook with the failure signature and rollback command so responders need not rediscover steps during an incident.

Observability

OpenTelemetry trace context injected in event metadata (traceparent). CloudWatch metrics on age of oldest message. Correlate business IDs across logs—not just request IDs from single Lambda.

Validate this in staging with production-like data volume before declaring done. Capture metrics baseline the week before change and compare for seven days after—subtle regressions hide in aggregates until a large tenant hits the path. Update the on-call runbook with the failure signature and rollback command so responders need not rediscover steps during an incident.

Choreographed saga: PaymentFailed event triggers inventory release. Orchestrated: Step Functions Catch invokes refund Lambda. Document failure matrix in runbook.

Validate this in staging with production-like data volume before declaring done. Capture metrics baseline the week before change and compare for seven days after—subtle regressions hide in aggregates until a large tenant hits the path. Update the on-call runbook with the failure signature and rollback command so responders need not rediscover steps during an incident.

Per-invocation billing favors batching SQS messages (batch size 10). EventBridge pricing per event—avoid chatty fine-grained events when one summary suffices.

Schema registry validates event compatibility on publish—breaking changes fail CI before production consumers break.

DLQ depth alerts with replay runbook after fix. Idempotency table TTL exceeds max redelivery window.

Choreography versus Step Functions: document failure matrix in runbook when payment fails mid-saga. Cost: batch SQS where per-message billing adds up.

Validate this in staging with production-like data volume before declaring done. Capture metrics baseline the week before change and compare for seven days after—subtle regressions hide in aggregates until a large tenant hits the path. Update the on-call runbook with the failure signature and rollback command so responders need not rediscover steps during an incident.

Document the decision, owner, and rollback path in your team wiki the same week you ship. Future you will not remember which environment variable toggled the behavior unless it is written next to the runbook entry and linked from the alert. That habit costs ten minutes per change and saves hours when pagination or auth misbehaves under a single large tenant.

Run the change through your standard PR checklist: tests, observability, and a two-minute rollback drill in staging. Small operational habits accumulate into systems that survive on-call nights without heroics.

Share a short write-up in your engineering channel after rollout: what shipped, what metric you watch, and who owns follow-up. That closes the loop for teammates who were not in the PR and surfaces gaps in docs before the next person repeats the same investigation.

Resources

• Amazon EventBridge documentation
• AWS SQS with Lambda
• Serverless Land event-driven patterns
• Enterprise Integration Patterns
• AWS Step Functions documentation