Skip to content

SLO Management

Service Level Objectives (SLOs) let you define reliability targets for your catalog components and track error budget consumption over time. Farm calculates budget status automatically from Prometheus metrics (or simulated data when Prometheus is unreachable or returns an empty result).

Core Concepts

Service Level Objective

An SLO sets a measurable target for a specific metric type over a rolling time window. For example, "API gateway must maintain 99.95% availability over 30 days."

Error Budget

The error budget is the acceptable amount of unreliability derived from the SLO target. A 99.95% availability target yields a 0.05 percentage-point error budget. Farm tracks how much of this budget has been consumed and alerts when consumption accelerates.

Burn Rate

Burn rate measures how fast the error budget is being consumed relative to the elapsed window. A burn rate of 1.0 means the budget is being consumed at a sustainable pace; values above 1.0 indicate the budget will be exhausted before the window ends.

Managing SLOs

Navigate to SLOs in the sidebar to see all configured objectives.

Creating an SLO

  1. Click Create SLO.
  2. Fill in the required fields:
Field Description
Name Unique identifier (e.g., api-availability)
Target % The reliability target (e.g., 99.95)
Metric Type availability, latency, or error_rate
Window Rolling evaluation window: 7d, 30d, or 90d
Description Optional human-readable explanation
Component Optional link to a catalog component
  1. Click Save. The SLO begins tracking immediately.

Editing an SLO

Click the pencil icon on any SLO card to modify its fields. Changes take effect on the next budget calculation.

Deleting an SLO

Click the trash icon and confirm. Historical budget data is not retained after deletion.

Error Budget Status

Each SLO card displays a status badge based on remaining budget:

Status Condition Color
Healthy More than 50% budget remaining Green
Warning 10-50% budget remaining Yellow
Critical Less than 10% budget remaining Red
Exhausted Budget fully consumed Red (solid)

Click View Budget on any SLO to see detailed metrics: current performance percentage, total budget, consumed budget, remaining budget percentage, and burn rate.

Metric Types

Type What it measures Prometheus query pattern (simplified)
availability Uptime percentage over the SLO window avg_over_time(up{job="<componentId>"}[<window>])
latency Response time compliance over the SLO window histogram_quantile(0.95, rate(http_request_duration_seconds_bucket{job="<componentId>"}[<window>]))
error_rate Error-free request ratio over the SLO window rate(http_requests_total{job="<componentId>",status!~"5.."}[<window>]) / rate(http_requests_total{job="<componentId>"}[<window>])

If Prometheus is unreachable or returns an error or empty result, Farm falls back to simulated metrics for development and demonstration purposes.

Best Practices

  • Start with availability SLOs for your most critical services before adding latency and error rate targets.
  • Use 30-day windows as the default. Seven-day windows are too noisy for most services; 90-day windows react too slowly.
  • Set realistic targets. A 99.99% target for a service that currently runs at 99.5% creates a permanently exhausted budget.
  • Link SLOs to catalog components so teams can see reliability targets alongside service metadata.