stardelt Services

The full set of services stardelt installs, split into the always-on core stack and the opt-in services that operators enable when they need them. For the architectural picture (what connects to what), see the Overview.

Core stack — always installed

The opinionated set every stardelt install ships. These services define what stardelt is.

stardelt Nova

Unified web UI: single SSO landing, catalog browser over Iceberg tables, lineage view, audit search, platform health, and deep-links into the native UI of every service below. TypeScript + React (front-end), Rust + axum (back-end). Apache 2.0.

• Repo: github.com/stardelt/stardelt-nova

stardelt Operator

The platform's CRD reconciler. Composes the services below from a small declarative API (PlatformInstance, Lakehouse, Pipeline, StreamApp). Rust + kube-rs. Apache 2.0.

• Repo: github.com/stardelt/stardelt-operator

Apache Trino

Distributed MPP SQL engine — fast interactive queries over Iceberg tables. Apache 2.0.

• Site: trino.io
• Repo: github.com/trinodb/trino

Apache Spark (Spark Connect)

Distributed compute engine. Deployed via the Spark Kubernetes Operator. Spark Connect exposes a persistent remote endpoint (sc://spark-connect-svc:15002) so notebooks and other clients connect without spawning their own driver. Apache 2.0.

• Site: spark.apache.org
• Spark Connect: docs
• Operator: github.com/kubeflow/spark-operator

Apache Airflow

Workflow orchestration for batch and scheduled jobs. KubernetesExecutor under the hood, so each task runs as its own pod. Apache 2.0.

• Site: airflow.apache.org
• Repo: github.com/apache/airflow

JupyterHub

Multi-user notebook environment — the front door for interactive analysis. Notebooks reach Trino directly (Python client), Spark heavy-lifting via Spark Connect, and (when enabled) Airflow via its REST API to promote interactive work into scheduled jobs. BSD-3.

• Site: jupyter.org/hub
• Repo: github.com/jupyterhub/jupyterhub

Apache Superset

BI and dashboarding over Trino. Apache 2.0.

• Site: superset.apache.org
• Repo: github.com/apache/superset

Apache Iceberg

Open table format on object storage. The shared storage abstraction every engine reads and writes. Apache 2.0.

• Site: iceberg.apache.org
• Repo: github.com/apache/iceberg

Lakekeeper

Iceberg REST catalog implementation in Rust. Vends table metadata and short-lived object-store credentials to engines. Apache 2.0.

• Site: lakekeeper.io
• Repo: github.com/lakekeeper/lakekeeper

SeaweedFS

S3-compatible distributed object store; the default data plane for Iceberg files. Operators may BYO any S3-compatible storage (AWS S3, GCS via S3 interop, Azure Blob, Ceph, NetApp) instead. Apache 2.0.

• Repo: github.com/seaweedfs/seaweedfs

Apache Kafka (KRaft)

Event-streaming backbone. Used for ingest, CDC events, and the platform's audit-event topic. KRaft mode (no ZooKeeper). Deployed via the Strimzi operator. Apache 2.0.

• Site: kafka.apache.org
• Strimzi: strimzi.io

Opt-in services — enabled per operator choice

These extend stardelt but are not required. Operators often have existing equivalents already in the cluster; stardelt installs these only when asked.

Keycloak

OIDC / SAML / LDAP identity provider — federates an upstream corporate IdP into stardelt. Skip if the cluster already has an OIDC provider. Apache 2.0.

• Site: keycloak.org
• Repo: github.com/keycloak/keycloak

Prometheus + Grafana

Metrics collection and dashboards. Skip if the cluster already has a monitoring stack. Apache 2.0.

• Prometheus: prometheus.io
• Grafana: github.com/grafana/grafana

Alertmanager

Alert routing and grouping, paired with Prometheus. Apache 2.0.

• Repo: github.com/prometheus/alertmanager

cert-manager

TLS certificate lifecycle for in-cluster and external endpoints. Apache 2.0.

• Site: cert-manager.io
• Repo: github.com/cert-manager/cert-manager

Argo CI

GitOps delivery (Argo CD) and pipelines (Argo Workflows) — drives stardelt installations from a Git repository instead of out-of-band kubectl. Apache 2.0.

• Site: argoproj.github.io
• Repo: github.com/argoproj

Apache SeaTunnel

Data integration / EL connectors — pulls data from operational systems (databases, queues, SaaS APIs) into Iceberg. Apache 2.0.

• Site: seatunnel.apache.org
• Repo: github.com/apache/seatunnel

Apache Flink

Stream processing — event-driven pipelines, CDC, real-time transformations off Kafka topics. Apache 2.0.

• Site: flink.apache.org
• Repo: github.com/apache/flink

Open Policy Agent (OPA)

Fine-grained policy enforcement for data and platform resources. Apache 2.0.

• Site: openpolicyagent.org
• Repo: github.com/open-policy-agent/opa

Service selection criteria

stardelt picks services by:

1. License: OSI permissive (Apache 2.0 / MIT / BSD). No BSL / SSPL / ELv2 / AGPL.
2. Production maturity in 2026: real users, active commits, recent releases.
3. Kubernetes nativeness: clean K8s deployment, ideally with a maintained operator.
4. Modern architecture: 2020+ design choices, not Hadoop/HDFS heritage.
5. Sovereignty compatibility: no required outbound calls at runtime.
6. Integration with the rest of the stack: e.g., Lakekeeper + Iceberg + Trino is a known-good combination.

When two candidates score equally, stardelt prefers CNCF or ASF projects over vendor-led ones, Rust or Go services over JVM where K8s controller-pattern fits, and projects with cloud-agnostic posture over those tightly coupled to one hyperscaler.