SLOs

Service Level Objectives define measurable commitments for ΔOS performance and reliability.

Availability SLOs

Intent Evaluation Availability

What this means: In a 30-day period, you can expect less than 22 minutes of evaluation unavailability.

Audit Recording Availability

What this means: Audit records are nearly always recorded synchronously. In rare cases, recording may be delayed but data is never lost.

Latency SLOs

Intent Evaluation Latency

Measurement: Time from Intent submission to Judgment return.

Exclusions:

• External evidence collection (customer-provided sources)
• Escalation routing (by design, waits for human)

Evidence Collection Latency

Throughput SLOs

Intent Processing Rate

Burst handling: Systems can handle 5x sustained rate for 60-second bursts.

Escalation Processing

Data Durability SLOs

Audit Trail Durability

What this means: Audit records are effectively permanent. You will not lose audit data.

Intent Data Durability

Consistency SLOs

Judgment Consistency

What this means: If you replay a historical Intent with its original evidence, you will get the same Judgment. Always.

Policy Propagation

Escalation SLOs

Routing Performance

SLA Tracking

Monitoring

Real-Time Metrics

Access current SLO status:

const sloStatus = await deltaos.slos.current();

console.log(sloStatus);
// {
//   evaluation: {
//     availability: { current: 0.9998, target: 0.9995, status: 'healthy' },
//     latencyP99: { current: '45ms', target: '100ms', status: 'healthy' }
//   },
//   audit: {
//     availability: { current: 0.99995, target: 0.9999, status: 'healthy' },
//     durability: { current: 1.0, target: 0.99999999999, status: 'healthy' }
//   }
// }

Historical SLO Data

const sloHistory = await deltaos.slos.history({
  metric: 'evaluation.latencyP99',
  period: '30d',
  granularity: '1h'
});

// Returns hourly P99 latency for the past 30 days

Status Page

Current SLO status is always available at status.deltaos.ai.

SLO Breaches

What Happens

When an SLO is breached:

1. Detection — Automated monitoring detects the breach
2. Alert — Engineering is notified immediately
3. Response — Incident response begins
4. Communication — Customers are notified via status page
5. Resolution — Issue is resolved
6. Review — Post-incident review conducted

Error Budgets

Customer Notification

// Configure breach notifications
await deltaos.notifications.configure({
  sloBreaches: {
    channels: ['email:ops@company.com', 'slack:#alerts'],
    severity: ['critical', 'warning']
  }
});

SLO Reporting

Monthly Reports

Enterprise customers receive monthly SLO reports:

• Actual vs target for all SLOs
• Incident summary
• Trend analysis
• Capacity projections

Export Data

const report = await deltaos.slos.report({
  period: '2025-01',
  format: 'pdf'  // or 'json', 'csv'
});

See Also

• Guarantees — Properties that cannot be violated
• Incident Posture — What happens when things fail
• Architecture: Components — System design