Skip to content
All case studies

Case study 03 · Logistics · E-commerce integrations

Resilient Salla & Zid integrations at Logestechs

Store orders flowing into shipment tracking and inventory — even when a third-party API is slow or down.

The problem

Merchants on Salla and Zid needed their orders to flow into the logistics platform for shipment tracking and inventory. Third-party APIs can be slow or briefly unavailable, and the integration had to keep working through that.

My role

I built the Spring Boot integration with Salla and Zid and the event-driven messaging around it, and tuned the high-traffic endpoints of the shipment and inventory services.

Architecture

My scopeDecisions, explained below
Architecture diagram: Resilient Salla & Zid integrations at LogestechsSalla and Zid stores send webhooks to the integration service, which also calls their APIs. The integration service passes orders to the shipment and inventory services, which use MySQL and a Redis cache. When a sync fails, it is queued in RabbitMQ and retried later.Salla · Zidstores · webhooks · APIsIntegration serviceSpring Boot · Resilience4jShipment & inventorySpring Boot servicesRabbitMQevents · retry queueMySQL + Redistuned queries · cachewebhooksAPI callsordersfailed syncsretry later

Key decisions

  1. Fail fast, recover gracefully. Every Salla and Zid call is wrapped in Resilience4j retries, a circuit breaker and a fallback, so a slow store API can't drag the platform down with it.

  2. Failed syncs aren't dropped. When a store API is down, the sync is queued in RabbitMQ and retried later instead of being lost.

  3. Event-driven between services. RabbitMQ messaging keeps the integration decoupled from the shipment and inventory services.

  4. Measure, then tune. SQL tuning, indexing and Redis caching on more than 10 high-traffic endpoints.

Result

lower API p95 latency
~15%
lower database load
~20%
high-traffic endpoints tuned
10+