Designing Scalable Backend Architecture with Microservices, Message Queues, and Scheduled Tasks

This article presents a three‑horse‑carriage architecture for internet projects, consisting of microservices, message queues, and scheduled tasks, which together form a stable, extensible backbone that rarely needs major redesign.

In the microservices section, the author stresses domain‑driven service decomposition, clear service boundaries, independent data stores, and layered service types (aggregation, basic, foundation). Benefits include easier requirement isolation, reduced code duplication, targeted scaling, and graceful decommissioning, while emphasizing proper granularity, idempotent APIs, and compensation mechanisms for cross‑service calls.

The message‑queue part outlines why MQs are introduced—async processing, handling traffic spikes, decoupling modules, and broadcasting events. Best practices include dedicated listener services, compensation jobs for lost messages, ensuring idempotency, monitoring queue health, and separating internal and external MQ clusters for security and performance.

For scheduled tasks, the author describes typical use cases, a task‑driven table schema (status, retry count, timestamps, etc.), and recommends using job frameworks such as ElasticJob or Quartz. Jobs should be isolated from business logic, focus on orchestration, and include compensation handling to avoid dead‑letter buildup.

The article then enumerates a concrete service breakdown (site, façade, business, foundation services, and job projects), showing how each module can be packaged independently, facilitating CI/CD while keeping the overall system coherent.

Overall, the proposed architecture balances extensibility and complexity, making it suitable for most internet applications while highlighting the importance of disciplined service design, monitoring, and incremental evolution.