Ontos

**Ontos** is a comprehensive Unity Catalog governance and metadata management application designed to run natively as a Databricks App. It empowers data teams to implement data mesh principles by enabling the creation, management, and discovery of Data Products, Data Contracts, and Business Glossaries—all within your Databricks environment.

**Key Benefits:**
- **Data Product Management**: Group and organize Databricks assets (tables, views, functions, models, dashboards) into discoverable data products with proper ownership and governance
- **Data Contracts**: Instrument data products with technical metadata following the Open Data Contract Standard (ODCS), including schema validation, quality checks, and access control
- **Business Glossary**: Maintain hierarchical glossaries that provide semantic context and enable consistent terminology across your organization
- **Role-Based Access Control**: Built-in RBAC with configurable personas (Admin, Data Producer, Data Consumer, Data Steward) to control who can create, edit, or consume data products
- **Compliance & Governance**: Automated compliance scoring and verification to ensure data products meet organizational standards

---

## **Use Cases**

- **Data Mesh Implementation**: Enable domain teams to publish and manage their data products independently while maintaining enterprise governance standards
- **Self-Service Data Discovery**: Provide a marketplace-like experience for data consumers to browse, subscribe to, and access certified data products
- **Data Contract-Driven Development**: Establish clear contracts between data producers and consumers before building pipelines, reducing integration issues
- **Business Glossary Management**: Create and maintain organization-wide or domain-specific business terminology to ensure consistent data interpretation
- **Compliance Auditing**: Continuously monitor and score data assets against defined compliance rules with automated notifications for violations
- **Entitlement Management**: Combine access privileges into personas and assign them to directory groups for streamlined permission management

---

## **Product Details**

Ontos manages and exposes metadata through the following core datasets:

- **Datasets represented include:** `data_products`, `data_contracts`, `datasets`, `business_glossary_terms`, `compliance_rules`, `entitlements`, `data_asset_reviews`, `settings`, `roles`, and `notifications`.

- **Sample fields include:** `name`, `description`, `domain`, `owner`, `version`, `status` (draft/active/deprecated), `schema_definition`, `quality_checks`, `access_permissions`, `compliance_score`, `created_at`, `updated_at`, and `tags`.

For more details, refer to the embedded notebook.

---

## **Additional Insights**

**Standards & Specifications:**
- Built on [BITOL Open Data Contract Standard (ODCS)](https://github.com/bitol-io/open-data-contract-standard) for data contract definitions
- Follows [BITOL Open Data Product Standard (ODPS)](https://github.com/bitol-io/open-data-product-standard) for data product specifications

**Architecture:**
- Runs as a native Databricks App with FastAPI backend and React/TypeScript frontend
- Stores metadata in Lakebase (Postgres) for high availability
- Supports Git sync for configuration version control
- Includes AI-powered "Ask Ontos" feature via LLM serving endpoints

**Resources:**
- GitHub Repository: [https://github.com/databrickslabs/ontos](https://github.com/databrickslabs/ontos)
- License: Databricks
- Version: 0.6.1

**Ontos** 是一款全面的**Unity Catalog**治理与元数据管理应用，可作为原生应用在Databricks平台上运行。它赋能数据团队落地数据网格（Data Mesh）理念，支持在Databricks环境内创建、管理和发现数据产品（Data Products）、数据契约（Data Contracts）以及业务术语表（Business Glossaries）。

**核心优势：**
- **数据产品管理**：将Databricks资产（表、视图、函数、模型、仪表盘）分组并组织为可发现的数据产品，并配置恰当的所有权与治理规则
- **数据契约**：遵循开放数据契约标准（Open Data Contract Standard, ODCS）为数据产品注入技术元数据，涵盖模式验证、质量检查与访问控制
- **业务术语表**：维护层级化术语表，为组织提供语义上下文并确保全企业术语使用一致
- **基于角色的访问控制（Role-Based Access Control，RBAC）**：内置可配置的角色体系，涵盖管理员、数据生产者、数据消费者、数据专员等角色，用于管控数据产品的创建、编辑与访问权限
- **合规与治理**：提供自动化合规评分与验证功能，确保数据产品符合企业标准

---

## **应用场景**
- **数据网格落地**：支持领域团队独立发布与管理其数据产品，同时维持企业级治理标准
- **自助式数据发现**：为数据消费者提供类数据集市的体验，使其可浏览、订阅并获取经过认证的数据产品
- **数据契约驱动开发**：在构建数据管道前，在数据生产者与消费者之间建立清晰的契约，减少集成问题
- **业务术语表管理**：创建并维护全企业或特定领域的业务术语，确保数据解读的一致性
- **合规审计**：依据预设合规规则持续监控并评分数据资产，对违规行为自动发送通知
- **权限管理**：将访问权限整合为角色，并分配至目录组，实现权限管理的流程简化

---

## **产品详情**
Ontos通过以下核心数据集管理并对外暴露元数据：
- **涵盖的数据集包括**：`"data_products"`（数据产品）、`"data_contracts"`（数据契约）、`"datasets"`（数据集）、`"business_glossary_terms"`（业务术语表条目）、`"compliance_rules"`（合规规则）、`"entitlements"`（权限配置）、`"data_asset_reviews"`（数据资产评审）、`"settings"`（设置）、`"roles"`（角色）以及`"notifications"`（通知）。
- **示例字段包括**：`"name"`（名称）、`"description"`（描述）、`"domain"`（领域）、`"owner"`（所有者）、`"version"`（版本）、`"status"`（状态，可选值：草稿/活跃/已弃用）、`"schema_definition"`（模式定义）、`"quality_checks"`（质量检查项）、`"access_permissions"`（访问权限）、`"compliance_score"`（合规评分）、`"created_at"`（创建时间）、`"updated_at"`（更新时间）以及`"tags"`（标签）。

如需了解更多详情，请参阅内置笔记本。

---

## **额外说明**
**标准与规范：**
- 基于[BITOL开放数据契约标准（ODCS）](https://github.com/bitol-io/open-data-contract-standard)定义数据契约
- 遵循[BITOL开放数据产品标准（ODPS）](https://github.com/bitol-io/open-data-product-standard)规范数据产品

**架构设计：**
- 作为原生Databricks应用运行，后端采用FastAPI，前端采用React/TypeScript构建
- 元数据存储于Lakebase（Postgres）以保障高可用性
- 支持Git同步以实现配置版本控制
- 集成基于大语言模型（LLM）服务端点的AI智能问答功能"Ask Ontos"

**资源信息：**
- GitHub仓库：[https://github.com/databrickslabs/ontos](https://github.com/databrickslabs/ontos)
- 授权协议：Databricks
- 版本：0.6.1

搜集汇总

背景与挑战

背景概述

Ontos是一个专为Databricks设计的Unity Catalog治理和元数据管理应用程序，支持数据网格原则的实施，包括数据产品、数据合同和业务术语表的管理。它提供了数据产品管理、数据合同、业务术语表维护、基于角色的访问控制以及合规性和治理等功能，旨在帮助组织实现数据治理和元数据管理。