惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

雷峰网
雷峰网
博客园 - 聂微东
酷 壳 – CoolShell
酷 壳 – CoolShell
宝玉的分享
宝玉的分享
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
罗磊的独立博客
Hugging Face - Blog
Hugging Face - Blog
T
Tailwind CSS Blog
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
IT之家
IT之家
博客园_首页
博客园 - 三生石上(FineUI控件)
博客园 - 叶小钗
Apple Machine Learning Research
Apple Machine Learning Research
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
量子位
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
人人都是产品经理
人人都是产品经理
美团技术团队
小众软件
小众软件
Jina AI
Jina AI
S
SegmentFault 最新的问题
博客园 - Franky
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
Choosing Your First SLI: A Decision Framework for New SRE...
Samson Tanimawo · 2026-06-21 · via DEV Community
Cover image for Choosing Your First SLI: A Decision Framework for New SRE Teams

Samson Tanimawo

New SRE teams ask the same question: what should we measure first? The temptation is to track everything CPU, memory, latency, error rate, queue depth, cache hit ratio. Don't.

You need one good SLI before you need ten mediocre ones.

What a first SLI should be

The first SLI you adopt should:

  1. Map directly to user experience. Not "CPU below 80%." Something more like "checkout requests succeed within 2 seconds." The closer to what users actually feel, the better.
  2. Be measurable from outside the system. If you can only measure it from inside (custom instrumentation in your code), you're going to fight that data forever. Start with HTTP status codes and request latency from a load balancer or CDN log.
  3. Have an obvious failure mode. "What does it mean when this metric drops?" If you can't answer in one sentence, pick a different metric.

The four candidates I'd consider first

For a typical web service:

  • Availability of the critical path. Successful HTTP responses (2xx, 3xx) divided by total requests, scoped to the routes that matter most. Not the whole API. Pick checkout, sign-in, search, whatever generates revenue or retention.
  • Latency at the 95th or 99th percentile. Median latency lies during incidents because half your users can still get through. Tail latency tells you when things break.
  • Job completion rate. If your system has background jobs (emails, reports, async exports), what fraction completes successfully within the time the user expects?
  • End-to-end synthetic checks. A canary that simulates a real user every minute. If it fails, you know something user-visible broke, even if internal metrics look fine.

Pick one. Just one. Get it stable for a month before adding a second.

What to skip until you're ready

  • Multi-burn-rate alerts. Powerful but complex. Most teams should start with a single threshold and learn what "normal" looks like before layering on math.
  • SLOs across multiple regions or services. Aggregating SLIs across regions is a science. Get one region right first.
  • Custom percentile math. Use whatever percentile your monitoring tool ships with. The difference between p99 calculated three different ways is rarely the gap between "ok" and "broken."

The trap to avoid

Teams over-engineer SLIs at the start because the literature is full of advanced patterns. Resist this. Pick a simple SLI, write down what "good" looks like, watch it for a quarter, then decide what's missing. The boring SLIs catch 80% of real problems for 10% of the effort.