Designing SaaS uptime like a reliability ledger

Ledgering uptime keeps owners accountable

We treat SLOs like a balance sheet. Every dependency gets a budget, every breach gets a root-cause memo, and we expose it to product so scope trades are explicit.

service: messaging-hub
slo:
  availability: 99.9
  latency_p95_ms: 600
dependencies:
  - name: sendgrid
    budget: 25%
  - name: auth0
    budget: 15%
alerts:
  burn_rate: 4h
  paging: squad-reliability

What makes a good ledger entry

Every dependency should have an owner, budget, and current burn rate. Capture noisy neighbors and vendor risk in the same view so product can trade scope with eyes open.

slo_ledger:
  service: api-gateway
  owner: platform
  budgets:
    auth0: 15%
    payments: 25%
    postgres: 35%
  alerts:
    burn_2h: page platform-oncall
    burn_24h: open ticket + slack #reliability
  runbook: https://runbooks.ndi/api-gateway-slo

What makes a good ledger entry

Every dependency should have an owner, budget, and current burn rate. Capture noisy neighbors and vendor risk in the same view so product can trade scope with eyes open.