SLO Creation, Error Budget Burn Rates & Alerting

Service Level Objectives (SLOs) and Error Budgets allow engineering teams to quantify service reliability, balance feature velocity against system stability, and trigger alerts before users experience major outages.

---

1. Core Concepts: SLIs, SLOs & Error Budgets

• Service Level Indicator (SLI): The actual measurement of service health (e.g., Good Requests / Total Requests or % of API responses < 200ms).
• Service Level Objective (SLO): The target reliability threshold over a rolling time window (e.g., 99.9% over 30 days).
• Error Budget: The allowable unreliability ($100% - 99.9% = 0.1%$).
• Burn Rate: The rate at which the error budget is consumed. A burn rate of 1x consumes 100% of the budget exactly at the end of 30 days; a burn rate of 14.4x burns 2% of the 30-day budget in 1 hour.

---

2. Step-by-Step: Creating an SLO in NudgeBee

1. Navigate to SLOs in the left navigation sidebar.
2. Click Create SLO.
3. Configure the general metadata:
   • Service Name: checkout-service
   • SLO Name: Checkout API 99.9% Availability
   • Rolling Window: 30 Days (or 7 Days, 90 Days)
   • Target Objective: 99.9%

---

3. Defining the Service Level Indicator (SLI)

NudgeBee supports two SLI evaluation models:

Model A: Ratio-Based PromQL Query (Recommended)

Provide separate queries for good events and total events:

• Good Events (Numerator):

  sum(rate(http_requests_total{job="checkout", status=~"2..|3.."}[5m]))

• Total Events (Denominator):

  sum(rate(http_requests_total{job="checkout"}[5m]))

Model B: Threshold-Based Metric

Direct latency percentage (e.g. histogram quantile under 200ms):

sum(rate(http_request_duration_seconds_bucket{le="0.2", job="checkout"}[5m]))
/
sum(rate(http_request_duration_seconds_count{job="checkout"}[5m]))

---

4. Multi-Window Multi-Burn-Rate Alerting

To eliminate false alarms from short transient spikes while promptly paging on severe outages, NudgeBee implements Google SRE Multi-Window Multi-Burn-Rate alerting:

Severity	Burn Rate	Short Window	Long Window	% Budget Consumed	Notification Action
Critical	14.4x	5 minutes	1 hour	2% in 1 hour	Delivery depends on configured notification rules
Critical	6x	30 minutes	6 hours	5% in 6 hours	Delivery depends on configured notification rules
Warning	3x	2 hours	24 hours	10% in 24 hours	Configure ticket creation separately if needed
Info	1x	6 hours	3 days	10% in 3 days	Delivery depends on configured notification rules

---

5. NuBi Documentation Search

Ask NuBi in chat for guided SLO and reliability assistance:

• "How do I configure a multi-window burn rate alert for a 99.9% availability SLO?"
• "What is the difference between ratio-based and threshold-based SLIs in NudgeBee?"