Skip to end of metadata
Go to start of metadata

ATLAS-1847 - Open Metadata and Governance Open


 

Open metadata and governance is a moon-shot type of project to create a set of open APIs, types and interchange protocols to allow all metadata repositories to share and exchange metadata.  From this common base, it adds governance, discovery and access frameworks to automate the collection, management and use of metadata across an enterprise.  The result is an enterprise catalog of data resources that are transparently assessed, governed and used in order to deliver maximum value to the enterprise.

Delivering this capability as open source is a critical part of the project since multiple vendors must buy into this ecosystem.  They are not going to do this if one organization dominates the technology base.  Thus the open metadata and governance technology must be freely available with an open source governance model that allows a community of organizations and practitioners to develop and evolve the base and then use it in their offerings and deployments.

The proposal is to use Apache Atlas as the open source reference implementation for open metadata and governance.  Apache Alas would support an open metadata and governance compliant repository plus provide the adapters and interchange formats to allow other metadata repositories to connect into the ecosystem.

This presentation provides an overview of the project: https://ibm.box.com/s/ofr6eelz39up4wkj9qlihsf55fsh2hft.

 


Project objectives

Figure 1 summarized the objectives of the new open metadata and governance function for Apache Atlas:  Automation, Business Value and Connectivity.

Figure 1: The ABC of open metadata and governance

  • Automation
    • Open metadata and governance provides an API for components that capture metadata from data platforms as data sources are created and changed.  This metadata is stored in the metadata repository and results in notifications to alert governance and discovery services about the new/changed data source.
    • It provides frameworks and servers to host bespoke components that automate the capture of detailed metadata and the actions necessary to govern data and its related assets.  This includes:
      • A discovery server that manages discovery services for analyzing and capturing metadata about new data sources.
      • A stewardship server that manages stewardship services for resolving issues detected during the governance and use of data.
  • Business Value
    • Open metadata and governance provides specialized access services and user interfaces for key data roles such as CDO, Data Scientist, Developer, DevOps Operator, Asset Owner, Applications.  This enables metadata to directly support the work of people in the organization.
    • The access services can also be used by tools from different vendors to deliver business value with open metadata. 
  • Connectivity
    • Finally connectivity enables a peer-to-peer Metadata Highway offering open metadata exchange, linking and federation between heterogeneous metadata repositories.

 

 

 


Technical components

The open metadata and governance project is divided into the following pieces:

  • Common types for open metadata - these types are built from the Apache Atlas type system and define the types stored in the graph database as well as payloads for notifications and APIs
  • Open Metadata Repository Services (OMRS) - Open metadata repository APIs and notifications to enable metadata repositories to exchange metadata in a peer-to-peer metadata repository cohort.  This capability is located in each metadata repository and collectively they enable what is referred to as the "metadata highway".
  • Open Metadata Access Services (OMAS) - Consumer-centric APIs and notifications for specific classes of tools and applications.  The OMAS services call the OMRS to access metadata from any open metadata repository.
  • New frameworks to complement the Atlas Hooks and Bridges
    • Open Connector Framework (OCF) - provides factories for connectors with access APIs for data resources and metadata together.  The OMRS is also built as a set of metadata repository connectors and the OMAS services use the OCF to connect to the appropriate OMRS connector.
    • Open Discovery Framework (ODF) - provides management for automated processes and analytics to analyze the content of data resources and update the metadata about them.
    • Governance Action Framework (GAF) - provides audit logging and governance enforcement services for implementing enforcement points in data engines, security managers such as Apache Ranger, and APIs.  It also adds stewardship services for analyzing audit logs and resolving issues identified in exceptions raised by the enforcement services.
  • Open Metadata Graph Repository - A set of stores linked together with a graph database.  These stores provide linkage between business, technical and operational metadata along with logs for auditing, operational lineage, metering and exception management..
  • Open Lineage Services - Services for collecting and querying lineage information across multiple heterogeneous metadata repositories

 

Figure 2 shows how this could look if it were implemented in Apache Atlas.

 

 

Figure 2: Open Metadata and Governance Components implemented in Apache Atlas

 

At this current time, there is a huge investment into Apache Atlas to add the open metadata and governance features plus also work on adoption across the data industry.  The wiki pages about this project represent the current stage of the architecture and design.  The JIRAs that are linked to are used for managing the code delivery.  Typically the wiki pages are ahead of the code since they are used to communicate the design as part of the code development cycle.

 


Integrating into the Open Metadata and Governance Ecosystem

With these frameworks and APIs in place, Apache Atlas becomes the reference implementation for the open metadata and governance APIs as well as offering the integration capability for the open metadata repository cohort.   Its function is divided into different packages to allow technology partners to connect into the open metadata and governance ecosystem.   The integration options are described as 4 patterns (native, caller, adapter and plug-in) here.