Skip to content

Kauzas Data Platform

You set the physical boundaries of your data, too.

Place your data in Türkiye.

With Kauzas Data Platform, place your enterprise data infrastructure on Kubernetes — in your own data center, in a private cloud or in AWS Local Zone İstanbul. Run the lakehouse, query, data processing and governance components in one architecture. You decide where the data is stored and processed.

  • Kubernetes-based
  • Built on open technologies
  • Runs on a closed network

Data sovereignty

Details

Where data is processed matters as much as where it is stored.

In modern data platforms, data is not merely stored — it is queried, processed and served to applications. Real data sovereignty therefore requires that the query, processing and serving layers stay under the organization's control alongside the storage layer.

Central SaaS architecture

  1. 01Part of the query may be planned on the provider's infrastructure.
  2. 02Query results may pass through the provider's serving layers.
Control boundary
  1. 03Data or query results may leave the organization's control domain.

The data path is not fully under the organization's control.

Kauzas

  1. 01The query runs on the organization's own infrastructure.
  2. 02Results pass through the organization's own serving layer.
  3. 03Data reaches only the authorized user, over a data path the organization controls.
Control boundary

The data path stays under the organization's control.

With Kauzas, data sovereignty is addressed not only at the storage layer but across the whole data flow.

Layered stack

What makes the difference is how the technologies work together.

Kauzas brings Iceberg, Trino, Spark and other open technologies together with shared management, authorisation and observability. Components can be brought up as needed and scaled independently, which reduces dependence on any single technology or vendor.

  1. L1

    Control plane

    User interface, API and orchestration

    • L2

      Serving and query

      Interactive SQL, hot serving, batch

      • Trino — federation
      • ClickHouse — hot tier
      • Spark on K8s — batch
    • L3

      Ingestion and pipelines

      Data in, transform, schedule

      • NiFi · Debezium
      • Airflow
      • dbt
    • L4

      Catalog and governance

      Technical catalog, discovery, lineage

      • Iceberg REST → Polaris
      • OpenMetadata
      • OpenLineage
    • L5

      Table format

      Table semantics

      • Iceberg — native
      • Delta · Hudi — interop
      • Hive — compatibility
    • L6

      Storage

      Physical bytes

      • S3 API — one interface
      • External connections — PostgreSQL · SQL Server
      • NAS — via gateway

    Products

    Infrastructure grows with the need, and carries nothing beyond it.

    Kauzas starts with the management layer alone. Platform services are brought online as they are needed, so the organization allocates resources only for the components it actually uses — and can scale and manage each one independently.

    Platform products and user resources

    Platform products

    The base component an administrator adds to the platform and manages centrally. It is installed, configured, updated and switched off when it is no longer needed.

    User resources

    Working resources that users create on the platform. A SQL warehouse, a notebook or a Spark job are examples.

    Product catalogueInteractive example

    13/13 products installed

    kauzas core

    control plane · always installed

    • Console
    • API / BFF
    • Cedar
    • Operator
    • State database
    Kauzas AI PlatformWorks with existing data lake and lakehouse infrastructuresExplore

    The components you select are brought up wired to work together. Each one can be versioned, deployed, scaled and switched off on its own.

    The console

    The whole platform is managed from one place.

    The Kauzas console brings every platform operation into a single interface — from putting products into service to running queries, from authorization to managing compute resources. An administrator brings up the components they need from here; prerequisite technologies, recommended components and install state are shown plainly on every card.

    kauzas.kurum.local
    Products

    Manage the components together, not one by one.

    Kauzas does not merely switch on the components you select; it also manages the dependencies and integrations between them. Governance, query engines, orchestration and the other services are wired to one another as they are needed, and both install-time and runtime relationships are handled by the platform. Each component therefore stops being a service that runs on its own and becomes a managed, interoperating layer of the platform.

    Trust

    Auditability is part of the architecture.

    In Kauzas, authorization changes, product installs and system activity are all recorded. Every action is traceable and verifiable, so the organization can see who did what, and when.

    Traceable audit chain

    This interactive example shows how a change to linked audit records can be detected by an integrity check.

    Traceable audit chain

    Chain intact

    Click a row — every hash after it becomes invalid.

    Authorization

    Authorisation through central policies, not limited to roles.

    In Kauzas, access rights are not defined by fixed roles baked into application code. Who may reach which data, product or compute resource, and under what conditions, is decided by central policies. Permissions can therefore be managed, audited and defined in fine detail without a code change.

    cedar policy

    permit(
      principal in Group::"data-engineers",
      action  == Action::"UseCompute",
      resource
    )
    when { resource.access == "public" };

    Authorization is enforced in the foundation of the system, not in the interface.

    Access controls do not stop at hiding screens; they are enforced on the server. A user can reach only the data and resources they are authorized for.

    Security

    Sensitive data is never stored in plain text.

    Connection details, access keys and SSO secrets are stored encrypted. Sensitive values are handled securely in both development and production, and are never written in the clear into application code or the database.

    Closed network

    It can run with no internet access.

    Kauzas can be installed and operated on closed networks with no external internet access. The packages and components it needs come from the organization's own controlled sources, and it depends on no external service at runtime.

    Licensing

    Component licensing terms

    Kauzas brings together components under different open source licences. The effect of the versions used and of the distribution method on licence terms has to be assessed separately.

    Management

    One platform, consistent management.

    Infrastructure changes go through GitOps, while day-to-day operations are handled from the Kauzas console. Both paths resolve to the same platform state, so configuration stays consistent, traceable and reversible when needed.

    Data integrity

    Every kind of data lives in the layer it belongs to.

    Configuration lives in Git, application state in the database, runtime in Kubernetes, and durable data in object storage. Every piece of information therefore has one definite source, and restarting a service leaves durable data untouched.

    Identity

    Enterprise identity management, resilient to outages.

    Kauzas connects to the organization's existing identity provider over OIDC and maps user groups onto platform permissions. When the provider is unreachable, controlled local administrator accounts keep operations running.

    Runtime environment

    The infrastructure can change; the platform does not.

    Kauzas runs the same Kubernetes architecture on AWS, on VMware and on physical on-prem infrastructure. The platform's operating model therefore does not change with the infrastructure: the organization picks the environment that suits it and stays free of any single cloud provider.

    What changes is the infrastructure; the platform architecture stays the same. Karpenter on AWS, Cluster API on VMware, and the existing server pool on physical infrastructure.

    Scaling per infrastructure

    • AWS

      Karpenter

      Kubernetes nodes are brought up as demand requires, and capacity is reduced automatically when unused.

    • VMware

      Cluster API (CAPV)

      Virtual machines are provisioned and managed automatically as Kubernetes nodes.

    • Physical infrastructure

      Existing server pool

      Kubernetes runs on the servers already in place; workloads scale at the pod level as needed.

    Operating model

    • A

      Fully in-house

      Default

      For environments that demand the highest level of control and data sovereignty. Both the management and the data layers run entirely on the organization's own infrastructure.

      Control plane
      In-house
      Data plane
      In-house
    • B

      Managed control plane

      Novadsa operates the Kauzas management layer while data and workloads stay on the organisation's own infrastructure — managed-service convenience together with control of the data.

      Control plane
      Novadsa / Kauzas
      Data plane
      In-house
    • C

      Central management, distributed sites

      For organizations with several locations or facilities: management is centralized while data and workloads stay at the sites where they live.

      Control plane
      Central in-house infrastructure
      Data plane
      At the organization's sites

    Get in touch

    Would you like to see how Kauzas would work in your organization?

    Let's review your infrastructure, data needs and use cases together. We'll demonstrate an architecture suited to your organisation and show where Kauzas can add value.

    Contact us directly:info@novadsa.com

    Contact

    The information you send is processed as described in the contact form privacy notice. Privacy notice