Google’s Site Reliability Engineering (SRE) handbook is a bible for modern operations. At the heart of SRE philosophy are the Four Golden Signals: Latency, Traffic, Errors, and Saturation.
While these signals were designed for massive planetary-scale infrastructure, they are incredibly useful for small SaaS teams. Let's see how a small engineering team can implement them without needing a full-time ops department.
The Four Golden Signals, Simplified
1. Latency
- What it is: The time it takes to service a request.
- Small SaaS translation: How fast is your app loading?
- How to monitor: Use a simple frontend analytics tool or basic APM middleware to monitor average loading times and response delays. Focus on the 95th and 99th percentiles, rather than the average, to catch users experiencing extreme delays.
2. Traffic
- What it is: A measure of demand on your system.
- Small SaaS translation: How many HTTP requests or API calls are you getting per minute?
- How to monitor: Track daily active users and incoming requests. Spikes can indicate marketing success (or a DDoS attack), while sudden drops indicate a critical DNS or frontend outage.
3. Errors
- What it is: The rate of requests that fail.
- Small SaaS translation: Are your users seeing
500 Internal Server Errorscreens? - How to monitor: Integrate a standard error catcher (like Sentry or Bugsnag) and monitor your HTTP error rates. Keep error rates below 1% of total traffic.
4. Saturation
- What it is: A measure of system fullness.
- Small SaaS translation: Are your database connections or server memory limits maxed out?
- How to monitor: Set up basic CPU/Memory warnings on your hosting provider (like Vercel, Render, or AWS). If saturation crosses 80%, it's time to upgrade or optimize your code.
Conclusion
You don't need a dedicated DevOps engineer or complex Kubernetes clusters to run a reliable SaaS. Start with these simple, actionable signals, monitor key scheduled tasks, and you'll stay ahead of outages.