A backend microservice that aggregates event data from an external XML provider, caches it in Redis, and exposes a RESTful search API. Built as a portfolio project to demonstrate production-ready patterns in Java 25 and Spring Boot.
I designed this service to solve a common backend challenge: consuming unreliable external APIs while guaranteeing fast, available reads for end users. The focus is on resilient data synchronization, a manual cache-aside pattern, and a clean layered architecture — without reaching for heavyweight frameworks until they are warranted.
The service implements a dual-layer persistence strategy combining Redis for speed and PostgreSQL for reliability.
When the external provider is operational:
- A background sync task fetches data from the provider with retry logic
- Data is persisted to PostgreSQL for long-term storage
- Data is written to Redis with a configurable TTL
- API requests to
/api/v1/searchare served directly from cache
The synchronization runs on a fixed schedule, keeping data continuously fresh.
When the external provider experiences downtime:
- The sync task detects the failure (HTTP 5xx, timeout) after 5 retries
- The sync aborts cleanly — the read path is unaffected
- The service continues serving responses from Redis or rebuilds cache from PostgreSQL
- Zero downtime is maintained throughout the outage
| Component | Technology |
|---|---|
| Language | Java 25 |
| Framework | Spring Boot 3.5 |
| Database | PostgreSQL 16 |
| Cache | Redis 7.4 |
| Build | Maven |
| Containerization | Docker & Docker Compose |
| Observability | Micrometer Tracing + Logstash |
- Docker & Docker Compose
- Make
- Java 25+ (for local development)
- Maven 3.9+ (or use the included
mvnwwrapper)
make build
make runThe API will be available at http://localhost:8080.
# Start only infrastructure
make infra
# Run the application with the local profile
./mvnw spring-boot:run -Dspring-boot.run.profiles=localThe provider URL is externalized and can be overridden at runtime:
# Docker
PROVIDER_URL=https://your-provider.com/api/events make run
# Local
PROVIDER_URL=https://your-provider.com/api/events ./mvnw spring-boot:runSearch Events:
GET /api/v1/search?starts_at=2024-01-01T00:00:00&ends_at=2024-12-31T23:59:59
| Parameter | Required | Description |
|---|---|---|
starts_at |
Yes | ISO 8601 datetime — range start |
ends_at |
Yes | ISO 8601 datetime — range end |
Response: JSON array of events within the specified time range.
event-aggregator-service/
├── src/main/java/com/rubdev/eventsync/
│ ├── cache/ # Cache management (RedisTemplate-based cache-aside)
│ ├── config/ # Spring configuration & @ConfigurationProperties
│ ├── controller/ # REST API endpoints
│ ├── model/ # Domain entities & provider XML DTOs
│ ├── repository/ # Data access layer (Spring Data JPA)
│ ├── service/ # Business logic & orchestration
│ ├── task/ # Scheduled background sync
│ └── utils/ # Application constants
├── docs/
│ ├── adr/ # Architecture Decision Records
│ ├── CASE1.png # Architecture diagram — provider available
│ └── CASE2.png # Architecture diagram — provider unavailable
├── docker-compose.yml
├── Dockerfile
├── Makefile
└── pom.xml
mvn testThe test suite covers:
- Unit tests — service layer logic and cache manager behaviour
- Integration tests — repository operations
- API tests — endpoint validation and error handling
- Bundled mock provider for fully self-contained demo runs
- Circuit breaker if retry complexity grows beyond current scope (Resilience4j)
- Exponential backoff with jitter to reduce thundering-herd against recovering providers
- Indexed pagination on the search endpoint for large datasets
- Redis Cluster configuration for multi-region deployments
Rubén Juárez Pérez
Software Engineer — 2026.
MIT — see LICENSE