nw-production-readiness

# Production Readiness

## Monitoring and Observability

### Application Monitoring
- **Performance**: response time | throughput | latency percentiles (P50, P95, P99)
- **Resources**: CPU | memory | database connections | cache hit rates
- **Errors**: exception tracking | error rate trends | integration failure detection
- **Business**: KPI tracking | conversion funnels | feature usage | revenue impact

### Infrastructure Monitoring
Server/container health and resource utilization | Network performance and connectivity | Storage capacity and I/O performance | Security event detection.

### Alerting Tiers
| Tier | Condition | Response |
|------|-----------|----------|
| Page | Service down, data loss risk, security breach | Immediate response |
| Urgent | Error rate >2x baseline, latency SLA breach | Response within 15 min |
| Warning | Capacity >80%, error rate trending up | Response within 1 hour |
| Info | Deployment complete, metric threshold crossed | Review next business day |

## Operational Procedures

### Incident Response
1. Detect: automated alerting identifies issue
2. Triage: classify severity, assign responder
3. Communicate: notify stakeholders per severity level
4. Resolve: apply fix or rollback
5. Review: post-incident review within 48 hours
6. Improve: update runbooks and monitoring based on findings

### Maintenance Procedures
Regular update and patching schedule | Backup verification (test restores quarterly) | Security vulnerability scanning (automated, weekly) | Performance baseline recalibration (after major changes).

### Knowledge Transfer
Operational runbooks for common procedures | Architecture documentation with system diagrams | Deployment procedures and configuration management | Troubleshooting guides for known failure modes.

## Quality Gates for Production Readiness

Before declaring production-ready, all must pass:
- [ ] All acceptance tests passing
- [ ] Unit coverage meets project standard (default: >= 80%)
- [ ] Integration tests validated
- [ ] Performance validated under realistic load
- [ ] Security scan completed (0 critical, 0 high)
- [ ] Monitoring and alerting configured
- [ ] Logging structured and searchable
- [ ] Rollback procedure documented and tested
- [ ] Runbook created for operational procedures
- [ ] On-call team trained on new feature

For CI/CD architecture lessons and measurement coupling pitfalls, see `cicd-and-deployment` skill.