46 lines
1.6 KiB
Markdown
46 lines
1.6 KiB
Markdown
---
|
||
title: "Cost As Distributed Systems Bug"
|
||
type: concept
|
||
tags: [sre, finops, observability, reliability, cost-optimization]
|
||
last_updated: 2026-04-20
|
||
---
|
||
|
||
# Cost As Distributed Systems Bug
|
||
|
||
"成本是分布式系统的 bug"——成本异常（cost explosion）不仅是财务问题，更是一种可靠性问题。
|
||
|
||
## Core Thesis
|
||
成本突然增加往往预示着系统即将发生故障。**成本突增应该被视为告警信号**，触发故障调查而非仅财务审查。
|
||
|
||
## Why Cost Signals Matter
|
||
1. **资源泄漏的指示器**：内存泄漏、连接池耗尽往往表现为成本逐步上升
|
||
2. **异常流量的标志**：DDoS 或滥用可能导致成本爆炸
|
||
3. **配置错误**：错误的资源配置可能导致资源过度使用
|
||
4. **级联效应的前兆**：某个组件故障可能导致其他组件超负荷运转
|
||
|
||
## Alerting Strategy
|
||
```
|
||
IF cost_increase > threshold:
|
||
ALERT("Cost anomaly detected - investigate system health")
|
||
```
|
||
|
||
将成本监控集成到 SRE 的告警体系中，而非仅作为 FinOps 的事后分析。
|
||
|
||
## Key Principles
|
||
- **Cost as Signal**：将成本指标视为系统健康的信号
|
||
- **Proactive Monitoring**：在成本失控前设置告警
|
||
- **Correlation Analysis**：将成本变化与其他系统指标关联
|
||
|
||
## Relationship to FinOps
|
||
FinOps 不仅是成本优化工具，也是 SRE 的可靠性工具。成本可观测性（Cost Observability）是现代 SRE 实践的重要组成部分。
|
||
|
||
## Related Concepts
|
||
- [[Cost-Optimization]]
|
||
- [[Observability]]
|
||
- [[FinOps]]
|
||
- [[Distributed-Systems]]
|
||
- [[Reliability]]
|
||
|
||
## Source
|
||
- SRE Weekly Issue #513 — [[sre-weekly-issue-513]]
|