Alert Tuning

Alert Tuning

Apply data-driven threshold and window recommendations to reduce noise while preserving signal.

Opportunities

4

0 applied

Expected Reduction

-15 alerts/week

if all applied

Avg Noise Score

32

of candidates

Trigger Count (7d)

21

noise volume

Tuning Opportunities

Each suggestion includes the proposed change, rationale, and expected impact

ALERT-009Low

Host CPU > 90%

api-gateway

Current

Threshold

> 90%

Window

10m

Triggers (7d)

12

Noise Score

87

Suggested

Threshold

> 95%

Window

30m

Expected Impact

-10 alerts/week

CPU spikes during backups are transient. Tighter threshold and longer window filter expected noise.

> 90% · 10m> 95% · 30m
ALERT-010Low

Disk usage > 85%

api-gateway

Current

Threshold

> 85%

Window

5m

Triggers (7d)

4

Noise Score

21

Suggested

Threshold

> 90%

Window

15m

Expected Impact

-3 alerts/week

Disk usage growth is gradual. Higher threshold with sustained window avoids premature alerts.

> 85% · 5m> 90% · 15m
ALERT-007Medium

API Gateway p99 > 200ms

api-gateway

Current

Threshold

> 200ms

Window

5m

Triggers (7d)

2

Noise Score

8

Suggested

Threshold

> 250ms

Window

10m

Expected Impact

-1 alert/week

p99 latency has natural variance. Slightly higher threshold and longer window reduce false positives.

> 200ms · 5m> 250ms · 10m
ALERT-001High

Checkout 5xx > 1%

checkout-api

Current

Threshold

> 1%

Window

5m

Triggers (7d)

3

Noise Score

12

Suggested

Threshold

> 1.5%

Window

5m

Expected Impact

-1 alert/week

Current 1% threshold triggers during normal traffic jitter. 1.5% retains sensitivity to real outages.

> 1% · 5m> 1.5% · 5m

Tuning Summary

Impact of applied changes

Applied

0/4

Alerts Eliminated (7d)

0

Noise Score Reduction

-0

Command Palette

Search for a command to run...