# Applications Catalog

Phoeniqs runs a set of applications and operators cluster-wide on our shared clusters, so you can use them directly from your tenant namespace without installing anything yourself. This page is the authoritative catalog of what is available on the shared clusters, what we manage for you, what is planned, and how to consume each item.

If an application you need is not on this list, you can request it from the platform team or, if you need full flexibility, provision a Hosted Cluster where you manage operators yourself.


# How consumption works on the shared clusters

Tenant namespaces are isolated and managed by policy. You cannot install operators yourself: OLM installs (OperatorGroup, Subscription, CatalogSource, InstallPlan) are rejected in tenant namespaces, and you cannot create cluster-scoped objects (CRDs, ClusterRoles, ClusterRoleBindings). Instead, you use the operators we run cluster-wide by creating their custom resources in your own namespace. The namespace admin role already has the permissions for them.


# Available, self-serve in your tenant namespace

The following applications/operators are available on the shared clusters. For each, create the corresponding custom resource in your namespace (see the "How" column). Cluster availability is noted per item.

Item Clusters How
CloudNativePG (PostgreSQL) basel, ai-2, ai-3 Cluster custom resource; backups through a Barman Cloud ObjectStore
ClickHouse basel ClickHouse operator custom resources (clickhouse.com)
S3 object storage (Ceph RGW) basel; ai-2/ai-3 reach it through the public basel endpoint Purchase in the portal, then an ObjectBucketClaim on ocs-storagecluster-ceph-rgw — see S3 Object Storage
Block / file storage basel: Ceph; ai-2/ai-3: IBM Spectrum Scale basel: ocs-storagecluster-ceph-rbd-encrypted (default), ocs-storagecluster-cephfs; ai-2/ai-3: ibm-spectrum-scale-fileset
Quay registry basel, ai-2, ai-3 QuayRegistry in your namespace — see Tenant Image Registry with Quay
Keycloak basel, ai-2, ai-3 Keycloak custom resource (k8s.keycloak.org); operator runs cluster-wide
OpenShift Virtualization basel, ai-2, ai-3 VirtualMachine custom resources; on basel use ocs-storagecluster-ceph-rbd-virtualization
cert-manager basel, ai-2, ai-3 Certificate / Issuer in your namespace
ingress-nginx basel, ai-2, ai-3 IngressClass external or internal; OpenShift Routes also work
KEDA (Custom Metrics Autoscaler) basel, ai-2, ai-3 ScaledObject / ScaledJob
Reloader / Reflector basel, ai-2, ai-3 Annotations on workloads / Secrets
Monitoring (Cluster Observability Operator) ai-2, ai-3 MonitoringStack, PersesDashboard
NVIDIA GPU basel, ai-2, ai-3 Request nvidia.com/gpu; GPU quota comes with your plan — see GPU Infrastructure
OpenShift AI / AI inference (KServe, LeaderWorkerSet) ai-2, ai-3 (OpenShift AI also on basel) Data Science Projects, InferenceService — see Access Your OpenShift AI
Confidential containers (Kata/CoCo) basel Kata runtimeClass — see Confidential Computing
Kuadrant (Red Hat Connectivity Link) ai-2, ai-3 Gateway API + AuthPolicy / RateLimitPolicy

# Platform-internal, not for client namespaces

The following components are managed by Phoeniqs for platform operations and are not for use in your tenant namespaces:

MetalLB, ExternalDNS (the multi-tenant instance watches only the multi-tenant namespace), OpenShift GitOps (Argo CD), Kyverno, Zabbix, Node Feature Discovery, ODF/NooBaa (NooBaa StorageClass is blocked by policy), ACM/MCE (HyperShift), MTV, OADP, etcd backup, and Flux.

Clients do GitOps through their tenant repository (Flux), not Argo CD.


# Planned / in progress

Item Status Notes
Kafka (Strimzi / AMQ Streams) planned Not offered yet; a legacy install exists outside the managed inventory and will be regularised first.
OpenSearch in progress An operator runs on basel but has known issues (shared CRDs, single cluster-wide instance); not offered until fixed.
Vertical Pod Autoscaler (VPA) in progress Merged in the platform config, not yet released to production.
Grafana operator in progress Routed to basel in the past; being brought under the managed inventory, not offered until then.
Kuadrant, Cluster Observability Operator, KServe/LWS on basel planned Basel is on the v2.x release line; these come with the move to v3.x.

# Consumption model

  • Operator we provide → self-serve: create the custom resources in your namespace.
  • Operator we don't provide → ask the platform team. We evaluate a cluster-wide install; CRDs are shared, so there is one version per cluster for everyone.
  • Your own or custom operators, a specific operator version, or cluster-admin → Hosted Cluster. See Hosted Cluster.

# Related Pages