Amazon EventBridge has rolled out an enhanced custom event bus purpose-built to eliminate the operational overhead of managing multi-account serverless architectures.
The Multi-Account Serverless Bottleneck
Organizations building event-driven applications on Amazon EventBridge typically start small: a single custom event bus residing in one account, owned by a single team. But as adoption scales across the enterprise, reality hits. AWS best practices mandate a multi-account structure for isolation and security, which forces teams to spin up complex webs of cross-account rules and bus-to-bus configurations.
Suddenly, the operational complexity that serverless was supposed to erase comes rushing back. Platform teams lose granular visibility into who is subscribing to what. Cross-account and bus-to-bus routing charges compound rapidly. Worse yet, teams requiring deterministic event ordering find themselves building brittle workarounds or abandoning the platform entirely for alternative tech stacks. Managing fragmentation across dozens of isolated buses becomes a full-time job for platform engineers.
Architecture of the Centralized Shared Bus
The newly introduced enhanced custom event bus changes this dynamic by letting platform teams deploy a single, centralized event bus shared across all AWS accounts within an AWS Organization.
Resource sharing is handled natively via AWS Resource Access Manager (AWS RAM). Publishers push events to the shared backbone without needing to know downstream consumers, while application teams independently create their own subscriptions. Platform engineers retain top-down visibility and fine-grained control over publisher and subscriber permissions without maintaining custom routing rules.
Ordering Guarantees and Synchronous Lambda Invocations
Asynchronous architectures thrive when event order does not matter, but domain realities—such as logistics tracking or financial ledgers—demand strict sequencing.
The enhanced bus addresses this by supporting ordered delivery alongside standard asynchronous flows on the exact same resource. Publishers attach an EventGroupId to payloads. EventBridge then delivers events sharing that ID in strict sequence to subscribers that opt into ordered delivery, leaving other consumers to process messages asynchronously.
To eliminate the latency and overhead of placing an Amazon SQS buffer between an event bus and AWS Lambda just to guarantee reliability, the enhanced bus introduces synchronous invocation targets. This mode confirms successful downstream processing before acknowledging the event back to the bus.
Unified Subscribers and Content-Based Deduplication
Managing event routing previously meant stitching together separate rules, targets, and retry policies across multiple AWS resources. The enhanced custom event bus introduces a unified Subscriber resource that consolidates event filtering, target configuration, retry policies, and dead-letter destinations into a single manageable unit.
Data integrity gets a massive upgrade via content-based deduplication.
For data transformation, subscribers can utilize JSONata expressions to reshape payloads on the fly before they hit downstream targets. EventBridge natively deserializes incoming Apache Avro or Protocol Buffers payloads into JSON, enabling fine-grained filtering across the full event body without forcing teams to write custom deserialization wrappers.
Throughput Economics and Availability
The financial model for event-driven architectures shifts alongside these architectural changes. The enhanced custom event bus abandons the traditional per-event fee structure—which heavily penalized multi-bus architectures with compounding cross-account tolls—in favor of an ingress and egress throughput pricing model. Publishers pay strictly for ingested volume, and subscribers pay for delivered volume.
Legacy custom event buses remain fully operational under the hood, relabeled as “Custom event bus – classic” with no forced migrations required.
