QuestionQ36

VerifiedApplying site reliability engineering practices

You need to define Service Level Objectives (SLOs) for a high-traffic, multi-region web application. Customers expect the application to always be available and to have fast response times, and they are currently satisfied with its performance and availability.

Based on current measurements over a 28-day window:

  • 90th percentile latency: 120ms
  • 95th percentile latency: 275ms

What latency SLO would you recommend the team publish?

Explanation

SLO targets should not simply mirror current measured performance, because doing so leaves no error budget — any minor regression would instantly breach the SLO. The recommended practice is to set the SLO slightly looser than the currently observed performance (while still meeting user expectations for a fast, reliable experience), giving the team room to absorb normal variability and future changes without violating the objective. Given current values of 120ms (p90) and 275ms (p275... p95), a target of 150ms (p90) and 300ms (p95) provides that reasonable buffer while remaining consistent with what customers already find satisfactory.

Community Discussion

No comments yet. Be the first to start the discussion!