Case Studies Highlights Summary

Every highlight bullet from the 7 pages in this section, gathered on one page and grouped by the page it came from.

How to Read a Case Study

• A case study documents a decision made under real constraints, not a template to copy verbatim
• The constraints section matters more than the resulting architecture diagram
• Case studies decay as your context diverges from theirs - check the date and the stack version first
• Metrics separate evidence from anecdote - adjectives without numbers aren't a case study, they're marketing

Reference: B2B SaaS API

• Multi-tenant B2B API with NestJS 11 on Fastify adapter
• Prisma + Postgres with tenant_id on every row
• Redis for cache, rate limits, and BullMQ backing
• OpenTelemetry traces from HTTP through queue publish
• K8s HPA on CPU and custom request latency metric

Reference: Worker Fleet

• Separate Deployment from API with shared job schema
• Idempotency keys on every side-effecting job
• DLQ with alert on depth > 0 for 15 minutes
• Concurrency tuned against Postgres pool max
• Graceful shutdown: close workers before exit

Reference: Internal CLI Tool

• ESM + TypeScript compiled to dist/ for Node 24
• Commander subcommands map to platform operations
• Published to private npm registry with semver
• Auth via OIDC device flow or env API token
• Integration tests invoke bin via node:test

Before/After: Express → Fastify Migration

• Throughput +35% and p95 -40% on JSON-heavy routes
• Dedicated migration sprint, not incremental strangler
• Breaking: middleware signature, res.json patterns, error handling
• Schema validation moved from Zod-only to Fastify compile
• Load test proved win before prod cutover

Before/After: Node 20 → 24 LTS Upgrade

• 12 services upgraded over 3 sprints with N-1 policy
• Native modules: bcrypt rebuild, sharp version bump
• OpenSSL 3 stricter cipher defaults broke legacy webhook client
• CI matrix caught issues before production
• engines field + Docker base image pin enforced compliance

Case Studies Best Practices

• Every reference cites Node LTS and framework versions
• Before/after includes metrics, not adjectives
• References are copy-paste starting points, not gospel
• Revisit case studies when major versions ship
• Link ADRs that motivated the architecture