Uptime SLA

Uptime targets, history и incident response

Q: Какой uptime SLA SLAtech?

Starter 99.5% monthly. Pro и Scale 99.9% monthly. Enterprise 99.95% monthly (single-tenant). Service credits apply per tier-specific credit policy. Status real-time на status.slatech.ai.

Q: Какой historical uptime за last 6 months?

Январь 99.98%, февраль 99.96%, март 99.92%, апрель 99.99%, май 99.94%, июнь (month-to-date) 99.97%. Rolling 6-month average: 99.96%. Два minor incidents (1 Qdrant index-rebuild, 1 Cloudflare edge partition) в течение этого period.

Q: Как измеряется uptime?

Synthetic-monitoring transactions из multiple AWS regions, hitting /v1/query/ask с known query, каждые 5 минут (Starter/Pro), каждую минуту (Scale), каждые 15 секунд (Enterprise). Включает API latency budget (response < 5 секунд = healthy). Independent verification доступен через status-page widget.

Q: Как incidents reported и resolved?

6-step procedure: detection (Sentry / synthetic) → triage (status page в течение 10 минут) → mitigation (runbook applied) → resolution → public post-mortem в течение 5 business days → customer follow-up с service credit. Enterprise tenants получают dedicated post-mortem call.

Q: Есть real-time uptime dashboard?

Да — status.slatech.ai surfaces 12 metric groups: p50 / p95 / p99 query latency, retrieval recall, channel-specific success rates, EMR / CRM webhook delivery rates, и more. Updated в real-time через live SSE feed.

Q: Что происходит во время security incident?

Та же procedure plus immediate Sentry security-incident channel notification, optional engagement legal counsel (privacy lawyers), и 24-hour customer notification SLA для significant incidents (per NIS2 alignment). Affected data, scope и mitigation enumerated в post-mortem.

Published SLA targets per tier, 6-month historical uptime metrics, service-credit policy и 6-step incident-response procedure. Real-time status: status.slatech.ai. Пара с security controls, architecture overview и compliance glossary.

SLA targets per tier

Starter (€39/мес)

Target месячного uptime 99.5%

Credit policy: Service credits не bundled. Best-effort SLA с email-channel support.

Monitoring: 5-минутный synthetic monitoring; status page real-time.

Pro (€89/мес)

Target месячного uptime 99.9%

Credit policy: 10% credit от monthly fee если uptime падает ниже 99.5%. 25% если ниже 99%.

Monitoring: 5-минутный synthetic monitoring + per-tenant query latency tracking; status page real-time.

Scale (€199/мес)

Target месячного uptime 99.9%

Credit policy: 10% credit от monthly fee если uptime падает ниже 99.5%. 25% если ниже 99%. 50% если ниже 95%.

Monitoring: 1-минутный synthetic monitoring + per-tenant query latency tracking; priority status notification.

Enterprise (custom)

Target месячного uptime 99.95% (single-tenant deployment)

Credit policy: Negotiated в MSA. Typical: 15% monthly credit per 0.1% below target. Direct CSM escalation channel.

Monitoring: 15-секундный synthetic monitoring; dedicated Pingdom dashboard; on-call SRE escalation procedure.

Historical uptime (last 6 months)

Месяц	Uptime	Incidents	Заметки
Июнь 2026 (м-т-д)	99.97%	0	Нет production incidents. WAF blocked 2.1M malicious requests.
Май 2026	99.94%	1	12-минутная degradation 2026-05-17 из-за Qdrant index-rebuild — resolved автоматически.
Апрель 2026	99.99%	0	Нет production incidents. Сильный месяц.
Март 2026	99.92%	1	8-минутный Cloudflare edge incident, affecting global edge routing — out of SLAtech direct control но всё ещё counted.
Февраль 2026	99.96%	0	Brief Azure region partition (West Europe ↔ North Europe) tested failover — no customer impact.
Январь 2026	99.98%	0	Stable month. Wedge upgrade в .NET 10 был zero-downtime.

Rolling 6-month average: 99.96%. Real-time live data: status.slatech.ai.

6-step incident-response procedure

1. Detection
Sentry alert или synthetic-monitoring failure триггерит PagerDuty-equivalent rotation. Acknowledged в течение 5 минут для P1, 15 минут для P2.
2. Triage
On-call SRE assesses impact scope (которые tenants, которые features). Status page updated в течение 10 минут от acknowledgment с initial impact assessment.
3. Mitigation
Standard runbook applied (failover, rollback, hotfix). Median mitigation time для known patterns: 15 минут. Novel incidents: 30-60 минут.
4. Resolution
Service confirmed restored across synthetic monitors и sample tenant transactions. Status page updated с resolution timestamp.
5. Post-mortem
Public post-mortem published в течение 5 business days. Includes timeline, root cause, customer impact (количество tenants affected, duration), corrective actions (technical + process).
6. Customer follow-up
Affected Pro/Scale/Enterprise tenants получают direct email с incident details + applicable service credit. Enterprise tenants получают dedicated post-mortem call.

Reporting security или uptime incident?

SECURITY incidents: [email protected]. UPTIME / customer impact: status.slatech.ai live feed.

[email protected] status.slatech.ai