AWS EventBridge's Enhanced Event Buses Add Ordering, Deduplication, and Org-Wide Sharing — What Changed

EventBridge classic event buses handle event routing at scale, but multi-account architectures built on them carry a specific pain: routing events between accounts means routing between buses, and each hop costs separately. Teams at enterprise scale have been working around this with fan-out patterns that add complexity without being able to eliminate the fundamental per-event-per-hop cost structure.

Enhanced custom event buses, now generally available in 13 regions, rearchitect the fundamentals in ways that matter specifically for teams running event-driven systems across AWS Organizations.

What’s Actually Different

Organization-wide sharing. A single enhanced event bus can now be shared across all AWS accounts in an organization. Instead of routing events through bus-to-bus connections with cross-account targets, publishers across the organization subscribe to a shared bus directly. This eliminates the routing topology complexity that has made multi-account event architectures progressively harder to manage as organizations grow.

Guaranteed event ordering. Publishers can include an EventGroupId in event metadata. EventBridge uses this to guarantee sequential delivery to subscribers that require ordered processing, while other subscribers continue receiving events asynchronously. The practical use case: a logistics system where downstream processing requires knowing that event B came after event A, without forcing all subscribers to accept ordered delivery when they don’t need it.

Content-based deduplication. Automatic duplicate detection within five-minute windows provides exactly-once semantics without requiring publishers to implement idempotency keys. This addresses one of the more tedious reliability requirements in event-driven architecture — ensuring that infrastructure retry behavior doesn’t cause downstream effects to execute multiple times.

Simplified Subscriber resource. Classic buses separate filtering, targeting, retry policies, and dead-letter configuration across multiple resource types. The new Subscriber resource combines all of these into a single manageable unit per subscriber. For organizations with many event consumers, this reduces the resource proliferation that makes classic bus infrastructure configurations difficult to audit.

JSONata payload transformation. Subscribers can reshape event payloads before delivery using JSONata expressions. This moves basic event transformation into the bus layer rather than requiring consumer services to handle format differences.

Avro and Protocol Buffers deserialization. Enhanced buses automatically deserialize binary-encoded events in both formats, enabling schema-registered filtering on structured binary payloads — relevant for teams whose event producers are already using binary serialization for throughput efficiency.

Pricing Change

The new pricing model uses ingress/egress throughput rather than per-event counting. In multi-bus architectures where events traversed several routing hops, each hop generated a separate per-event charge. The throughput model replaces this with entry and exit costs on the enhanced bus, eliminating the compounding effect of complex routing topologies.

For simple single-bus architectures, the pricing comparison will depend on volume and message size. For organizations running complex multi-account event routing, the throughput model will generally be more predictable and lower at scale.

The 10,000 Subscriber Default

The default quota of 10,000 subscribers per bus (with higher quotas available on request) is the number to note for enterprise architects. Classic bus architectures that have scaled by adding buses — partly because of the practical limits on targeting — can now consolidate toward fewer buses with higher subscriber counts, simplifying the topology and governance model.

What Doesn’t Change

Classic event buses continue to exist and are not being deprecated. The enhanced buses are a new resource type rather than an in-place upgrade. Teams migrating from classic to enhanced buses will do so by creating new enhanced buses and migrating publishers and subscribers — the migration is migration work, not a configuration change.

Source: AWS News Blog

Content created with AI assistance and reviewed for accuracy.