监控与告警
incident-runbook-performance
更新于 2026-09-12
Incident Runbook · 性能规则
Performance rules for Incident Runbook: baseline first, dominant-cost-first fixes and recorded budgets.
- 技术栈
- Incident Runbook
- 主题
- 性能
- 条目
- 9 条
- 版本
- v0.2.0
✓ 应当这么做
- ✓ Measure a baseline on realistic data before changing anything
- ✓ Fix the dominant cost first: network, then N+1 queries, then allocation
- ✓ Record the budget and the measurement method next to the code
✕ 常见反模式
- ✕ Optimising code paths not visible in the profile
- ✕ Micro-benchmarks that ignore cold start and contention
- ✕ Trading correctness for speed without an explicit decision record
◎ 交付前验证
- ◎ The new code path is visible in logs end-to-end
- ◎ Dashboard panel added or updated to match the new metric
- ◎ Alert routing tested once end-to-end
同分类条目
9 篇监控与告警
Monitoring & Alerts 工程规则
Alerting rules: symptom-based signals with an owner and a runbook
监控与告警
Structured Logging 工程规则
Logging rules: correlation ids, bounded fields and no personal payloads
监控与告警
Distributed Tracing 工程规则
Tracing rules: propagator consistency and cardinality-safe span names
监控与告警
Incident Runbook 工程规则
Runbook rules: detection, mitigation and evidence capture per incident class