Managed Cloud
Operations under an agreed SLA: observability, ITSM, backup and DR, security operations.
How a managed estate is run
Signals from every layer, alerts with a first action, tickets in your ITSM, and recovery that is drilled.
Point at a component to light its links; select it for what it does here and the practice that makes it right on AWS.
What the managed service covers
Eight areas. Customers choose the ones they need; the rest stay with their own teams.
Reference service levels
NGT's reference SLA: the starting point, agreed per customer.
| Priority | Typical case | Response | Resolution |
|---|---|---|---|
| P1 · Critical | Production down with no workaround, or financial closing blocked | 30 minutes | 4 hours |
| P2 · High | Production degraded with a workaround, or a critical non-production system down | 4 working hours | 1 working day |
| P3 · Medium | Non-critical degradation; one function failing | 1 business day | 3 business days |
| P4 · Low | Minimal impact, not time-sensitive | 3 business days | 5 business days |
ISO 27001, HIPAA/HITRUST, PCI DSS, GDPR, CIS benchmarks and OWASP: the customer's frameworks we implement and operate, not certifications held by NGT.
Observability: hear about incidents before users do
Open standards from the application to the dashboard, so the stack can change without losing history.
- 01InstrumentOpenTelemetry in the application and agents on hosts: metrics, logs and traces from one standard.OpenTelemetryAPMReal user monitoringSynthetics
- 02Collect and enrichA telemetry pipeline adds context and routes each signal to the right store.Collector pipelineEnrichmentMulti-tenant routing
- 03StoreMetrics, logs and traces in back ends built on object storage, queried together.PrometheusGrafana LokiGrafana TempoDatadogSplunk
- 04Show and alertRED dashboards per service; each alert routed with its first action written on it.GrafanaService dashboardsRunbook linksOn-call routing
Backup and DR, tested rather than assumed
A backup job that reports success is not a recovery plan. Every link in the chain is checked.
- Targets per systemRecovery point and time objectives come from the business owner, system by system.
- Verify the backup chainAlert on a broken chain, not only a failed job; restore, time it, record the gap.
- Drill in tiersTabletop, then an isolated restore in another region, then a zone drain; again after major changes.
- Close the gap in phasesA cost ladder from backup-only to warm standby; the owner picks the step that fits.
Point at a component to light its links; select it for what it does here and the practice that makes it right on AWS.