Customer SLA
Customer-facing service commitments against defined scope, service time and approved evidence.
ENTERPRISE
Observability with governance, continuity and service assurance.
AmrudAi is being designed around identity, policy, multi-tenancy, audit, controlled change, high availability, disaster recovery and reproducible service evidence.
Identity & RBACService AssuranceHA / DRAudit IntegrityPolicy-as-codeData PortabilityNotificationsControlled Change
SERVICE ASSURANCE
SLA, OLA, Provider SLA and SLOs from versioned evidence.
Service outcomes are intended to be reproducible from versioned scope, formula, calendars, exclusions and evidence.
Internal OLA
Supporting team obligations and handoffs segmented by effective ownership intervals.
Provider / Underpinning SLA
Upstream carrier, cloud or provider obligations without making external ticket state health authority.
SLI / SLO / Error Budget
Defined indicators, objectives, burn-rate evaluation and allowable budget where mathematically applicable.
OPEN→PRELIMINARY→REVIEW→FINAL↳RESTATED revision
IDENTITY & ACCESS
Role-aware operations across tenant and resource scope.
- Role templates plus custom roles
- SSO/MFA enterprise identity targets
- SCIM v2-compatible lifecycle direction
- Customer/MSP scope separation
- API service accounts under the same governance model
- Entitlement and authorization remain separate decisions
Platform Super AdminMSP Super AdminTenant AdminSecurity AdminNOC ManagerNOC L1Network L2/L3InfrastructureCloudDatabaseNCMIPAMAuditorExecutive Viewer
CONTINUITY
High availability, DR and buffered collection.
Exactly two Core nodes for HA
Active/Standby is the current v1 architecture target; ambiguous states should fail safe.
Sensors keep collecting
Collection and local buffering are designed to continue through transient Core/WAN interruption while capacity permits.
DR is separate from HA
Site-level disaster recovery is part of product scope; contractual RPO/RTO remain deployment/certification decisions.
Protect live monitoring
Storage protection is designed so retention pressure cannot silently consume resources needed for current monitoring.