Comprehensive Guide to Backend Architecture: Microservices, Service Mesh, Observability, and Messaging

Overview

The "Architect's Map" outlines a three‑part series focusing on backend architecture, emphasizing technology stacks, design principles, and scenario‑based decisions to guide optimal system design within constraints such as team experience, cost, resources, schedule, and business stage.

Microservices

Microservices are a modular architectural style that decomposes applications into small, independently deployable services communicating via language‑agnostic APIs. The article discusses theoretical foundations (Conway’s Law), domain‑driven design steps, and the importance of aligning service boundaries with business domains.

Gateway

Access gateway provides traffic entry, security, and global policies.

Business gateway (or BFF) handles routing, authentication, rate limiting, circuit breaking, service aggregation, and plugin extensions, with examples such as Kong, APISIX, Spring Cloud Gateway, and gRPC‑Gateway.

Service Registration & Discovery

Server‑side mode uses load balancers and DNS.

Client‑side mode relies on registration centers (etcd, ZooKeeper, Consul) for health checks and direct client load balancing.

Configuration Center

Beyond key‑value storage for registration, configuration centers (Nacos, Apollo) provide UI, permission management, and integration with Kubernetes ConfigMap/Secret for zero‑dependency deployments.

Observability

Observability consists of tracing, logging, and metrics. Tracing tools (Jaeger, Zipkin, SkyWalking) follow the OpenTracing standard, while OpenTelemetry unifies tracing, logging, and metrics collection.

Tracing Example

[Span A]  ←←←(the root span)
|
+------+------+
|          |
[Span B]      [Span C] ←←←(Span C 是 Span A 的孩子节点, ChildOf)
|          |
[Span D]      +---+-------+
|          |
[Span E]    [Span F] >>> [Span G] >>> [Span H]
↑
↑
↑
↑
（Span G在Span F后被调用，FollowsFrom）

Logging

Business logs follow RFC 5424 levels (Emergency to Debug). Centralized collection can use Filebeat, Fluentd, or Logstash, with storage in Elasticsearch and visualization via Kibana.

Metrics

Metrics are captured with Prometheus (counters, gauges, histograms, summaries) or Zabbix for traditional environments, enabling resource and business monitoring.

Service Mesh

Service mesh abstracts inter‑service communication, providing discovery, traffic control, security, and observability. Popular implementations include Istio (Envoy sidecar) and Linkerd, with considerations for high availability and sidecar deployment strategies.

Message Queues

The article compares Redis, RabbitMQ, Kafka, RocketMQ, and Pulsar across HA, throughput, feature richness, duplicate consumption, and ordering guarantees, highlighting each system’s trade‑offs and suitable use cases.

Conclusion

Choosing the right backend components requires balancing performance, reliability, and operational complexity, with a focus on modular design, observability, and appropriate messaging solutions.