Comprehensive Guide to Backend Architecture: Microservices, Service Mesh, Messaging, and Observability
This article provides a detailed overview of modern backend architecture, covering microservice fundamentals, design principles such as Conway's Law and DDD, gateway patterns, communication protocols, service registration, configuration management, observability pillars, service mesh options, and a comparative analysis of popular message‑queue technologies.
Overview – The "Architecture Map" concept is introduced as a macro‑level view of a backend architect’s knowledge, focusing on open‑source, cloud‑native, and third‑party services. The article is divided into three parts: microservices and messaging, databases and distributed systems, and DevOps/project management.
Microservices – Defined as a software style that builds applications from small, language‑independent building blocks communicating via APIs. Benefits include independent deployment, scalability, and reduced change impact. The theoretical foundation references Conway’s Law and Domain‑Driven Design (DDD) for service boundary decisions.
Design Principles – Conway’s Law: system design mirrors organizational communication structures. DDD steps: identify domains, bounded contexts, entities, value objects, aggregates, and repositories, then iterate the model in code.
Gateway – Described as the unified entry point handling traffic, security, and routing. Two layers are distinguished: access gateway (global traffic and security) and business gateway (authentication, rate limiting, circuit breaking, aggregation, plugins, BFF). Examples of technologies include Kong, APISIX, Spring Cloud Gateway, gRPC‑Gateway, Goku, Kratos, and go‑zero.
Communication Protocols – Comparison of HTTP REST (API style) and RPC (point‑to‑point) with examples such as Dubbo, Motan, Thrift, and gRPC, discussing their trade‑offs.
Service Registration & Discovery – Two modes: server‑side (load balancer + DNS) and client‑side (registration centre like etcd, ZooKeeper, Consul). The article includes a diagram of a tracing span hierarchy: [Span A] ←←←(the root span) | +------+------+ | | [Span B] [Span C] ←←←(Span C is child of Span A) | | [Span D] +---+-------+ | [Span E] [Span F] >>> [Span G] >>> [Span H] ↑ ↑ ↑
Configuration Center – Discusses KV stores (etcd, ZooKeeper, Consul) and cloud‑native solutions (Kubernetes ConfigMap/Secret, Nacos, Apollo) for storing service configs and enabling non‑developer access.
Observability – Covers the three pillars: tracing (OpenTracing, OpenCensus, OpenTelemetry; tools like Jaeger, Zipkin, SkyWalking), logging (log levels, collection via Filebeat, Fluentd, Logstash, storage in Elasticsearch/Kibana), and metrics (Prometheus counters, gauges, histograms, summaries; comparison with Zabbix).
Service Mesh – Explains the sidecar pattern and lists popular meshes (Istio, Linkerd, Conduit) with notes on high availability and deployment considerations.
Message Queues – Defines MQs, their role in decoupling, and evaluation criteria (HA, throughput, features, broadcasting, persistence, ordering). Provides a comparative table of Redis, RabbitMQ, Kafka, RocketMQ, and Pulsar, highlighting strengths and weaknesses of each.
References – Includes links to OpenTelemetry sampling processor, a detailed MQ comparison article, and the RocketMQ‑to‑Pulsar migration tool ROP.
IT Architects Alliance
Discussion and exchange on system, internet, large‑scale distributed, high‑availability, and high‑performance architectures, as well as big data, machine learning, AI, and architecture adjustments with internet technologies. Includes real‑world large‑scale architecture case studies. Open to architects who have ideas and enjoy sharing.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.