The problem
“We correct a product in one system, but the others keep the old data. Each team has a different version.”
Context
- Operation
- Retail chain, Nicaragua
- Systems involved
- Master catalog · service bus · consuming systems
The approach
The master system publishes an event when a product changes. The bus distributes it to each consumer, which adapts and applies the change to its own model.
Ordering by product
A version per product lets consumers discard old updates. The event contract defines which system owns each field.
Retries per consumer
Each subscription acknowledges its own work. An idempotency key prevents repeated effects when an event is delivered again.
When one system disconnects
Other consumers continue. The affected subscription retains pending events within the agreed retention period and supports replay.
Permissions and operations
Each consumer accesses only its own subscription. Retention, alerts and latency targets are agreed; connections and secrets are protected.
Outcome
- One update point for master data.
- Systems receive changes independently.
- Pending work visible per consumer.