Performance & reliability reporting

Single points of failure and under-provisioned resources, surfaced before they cause an incident.

High availability findings

Production resources with no failover configured — VMs outside an Availability Zone, RDS without Multi-AZ, single-replica Kubernetes Deployments — scored separately from security findings so reliability gaps don't get lost in a security backlog.

Rightsizing signals

CPU utilisation from CloudWatch, Azure Monitor, Cloud Monitoring, or Prometheus is reduced to P50/P95/P99 (not raw averages) to flag both under-provisioned resources (a performance risk) and over-provisioned ones (a cost opportunity) from the same data.