Architecture

1. Overview

What it is: Structural decisions about how components communicate, scale, and fail.

DevOps angle: Architecture determines deployability, observability, blast radius, and toil.

When to use: New systems, major rewrites, platform standards.

When NOT to: Microservices by default for a 3-person team — complexity is a cost.

2. Concepts

• Monolith — one deployable; simple ops early
• Microservices — independent deploy; distributed failure modes
• Modular monolith — boundaries without network
• Event-driven — async facts/events
• Message queues — RabbitMQ (commands/queues), Kafka (log/stream)
• API Gateway — north-south entry
• Service discovery — DNS, Consul, K8s Services
• CQRS / saga — advanced consistency patterns
• Twelve-Factor — config/logs/disposability

3. Architecture

graph TB
  Client --> GW[API Gateway]
  GW --> Auth[Auth Service]
  GW --> Ord[Order Service]
  Ord --> Q[Kafka / RabbitMQ]
  Q --> Pay[Payment Service]
  Q --> Email[Email Worker]
  Ord --> DB1[(Orders DB)]
  Pay --> DB2[(Payments DB)]

4. Production Best Practices

• Choose simplest architecture that meets scale/team needs
• Explicit contracts (OpenAPI/AsyncAPI); versioning
• Idempotent consumers; outbox pattern for reliable events
• Define failure domains and ownership
• Standardize platform (auth, telemetry, CI) so services don't reinvent
• Data ownership per service — no shared mutable DB across teams casually

5. Common Mistakes

Mistake	Impact	Fix
Distributed monolith	Worst of both	Real boundaries or merge
Chatty sync calls	Latency/cascades	Bulkhead + async
Shared DB	Coupling	Split ownership
No correlation IDs	Unoperable	Platform middleware

6. Debugging Guide

Map dependency graph; find timeout cascades; check queue lag; verify contract versions between producer/consumer.

7. Security

• Zero-trust service identity
• Gateway authn; service authz
• Least privilege data stores
• Threat model trust boundaries

8. Performance

• Avoid unnecessary hops
• Cache at right layer
• Partition Kafka by key carefully
• Backpressure for consumers

9. Real Production Example

Modular monolith initially; extract payment service when team/scale justified; Kafka for order events; Kong gateway; shared platform Helm library for telemetry/auth.

10. Interview Questions

Beginner: Monolith vs microservices?

Intermediate: When Kafka vs RabbitMQ?

Senior: Design migration from monolith without big-bang. Multi-region consistency.

11. Checklist

• Goals & constraints written
• Boundaries & data ownership clear
• Failure modes documented
• Observability standard
• Deploy/rollback story per component

12. Cheat Sheet

Style	Strength	Weakness
Monolith	Simple ops	Scale teams/deploy
Microservices	Independent deploy	Distributed complexity
Events	Decoupling	Eventual consistency

Related Skills

• cicd
• kubernetes
• kong
• observability
• databases

---

Author & maintainer: Shanmukha Kumar Karra
Created and maintained as part of AgentOps Kit.