# Platform layers
## Shared platform services
This architecture view connects the services that provide ingress, certificate
management, secret delivery, persistent storage, metrics, traces, and logs for
Kubernetes workloads.
Open the interactive shared-services architecture
The diagram reflects the service description and the current declarative
repositories. Two findings remain explicit rather than being inferred away:
- the service description names Loki but does not define the downstream log
backend topology;
- the current Fluent Bit repository credential points to the Cert Manager
repository URL and should be reviewed.
## Observability
The Prometheus production architecture covers metric discovery and scraping,
Prometheus persistence, Alertmanager routing, Grafana access and LDAP,
Vault-sourced secrets, and the additional Tempo, staging, and Kasten data
sources used by Grafana.
Open the interactive Prometheus production architecture
The production stack uses `kube-prometheus-stack` and Ceph RBD persistence for
Prometheus and Alertmanager.
## Storage
The production storage view follows persistent claims through the Ceph RBD and
NFS CSI drivers to their external data systems. It also shows the CSI snapshot
API, snapshot controller, Kasten-enabled snapshot classes, and Vault-backed
Ceph credentials.
Open the interactive production storage architecture
Ceph RBD is the default production class with `Delete` reclaim policy; NFS
provides retained shared storage.