Four concepts you will use in every interaction with DataHub:

• Entities
• Aspects
• URNs (Uniform Resource Names) — each entity's permanent ID
• Domains

• An entity is a specific "thing" that DataHub tracks -- think of it as one row in a directory
• DataHub tracks many kinds of entities; the major ones:
  • Dataset: table, view, topic, file, or any structured data asset
  • Dashboard: a Looker dashboard, Tableau workbook, etc.
  • Chart: an individual chart or visualization

• Every entity has a unique resource name (URN) -- its permanent ID, assigned when the entity is created and never changed
• Almost every entity also has a human-readable name -- the friendly label you see in the UI
• Example: the table you see as fact_revenue_daily is, under the hood, urn:li:dataset:(urn:li:dataPlatform:snowflake,PROD.FINANCE.FACT_REVENUE_DAILY,PROD)
• The URN looks scary, but you rarely type one -- the UI handles it; you just need to recognize one when you see it

• Entity type: the category (Dataset, Dashboard, User...)
• Entity: a specific instance of that type
• Dataset: urn:li:dataset:(urn:li:dataPlatform:snowflake,PROD.FINANCE.FACT_REVENUE_DAILY,PROD)
• Dashboard: urn:li:dashboard:(looker,dashboards.42)
• User: urn:li:corpuser:jsmith
• The confusion shows up in bug reports: "The entity is wrong" -- which entity, specifically?

• An aspect is a named bundle of properties attached to an entity
• One entity has many aspects -- each aspect holds one category of metadata
• Example: the FACT_REVENUE_DAILY Dataset entity has these aspects, among others:
• SchemaMetadata: column names, data types, field descriptions
• DatasetProfile: row counts, null counts, sample values
• Ownership: who owns this dataset (and in what role)
• GlobalTags: tags applied to this dataset

• Multiple systems can write different aspects to the same entity simultaneously
• Example flow for FACT_REVENUE_DAILY:
• Snowflake ingestion writes the SchemaMetadata aspect (columns, types)
• dbt ingestion writes the DatasetProperties aspect (description)
• Airflow integration writes the DataJobInputOutput aspect (pipeline linkage)
• Data steward (UI) writes the Ownership aspect (assigns an owner)
• Governance team writes the GlobalTags aspect (adds a PII tag)

• URN stands for Uniform Resource Name
• Every entity in DataHub has exactly one URN -- it never changes
• Format: urn:li:<entity-type>:(<identifier>)
• Snowflake table: urn:li:dataset:(urn:li:dataPlatform:snowflake,PROD.FINANCE.FACT_REVENUE_DAILY,PROD)
• Looker dashboard: urn:li:dashboard:(looker,dashboards.42)
• Individual user: urn:li:corpuser:jsmith

• Use URNs, not display names, whenever referencing an asset programmatically
• Display names can change. URNs cannot.
• In API calls: pass the full URN as the entity identifier
• In bug reports or support tickets: include the URN so the exact asset is unambiguous
• In scripts: resolve the display name to a URN once at the start, then use the URN throughout
• Where to find a URN: the entity's URL in the DataHub UI contains it (URL-encoded)

• When you navigate to any entity in the DataHub UI, the URN is in the URL
• Example URL: https://your-datahub.example.com/dataset/urn%3Ali%3Adataset%3A%28urn%3Ali%3AdataPlatform%3Asnowflake%2CPROD.FINANCE.FACT_REVENUE_DAILY%2CPROD%29
• URL-decoded, the path after /dataset/ is: urn:li:dataset:(urn:li:dataPlatform:snowflake,PROD.FINANCE.FACT_REVENUE_DAILY,PROD)

• A domain is a named business grouping that organizes assets in DataHub
• Example domains: Finance, Marketing, Engineering, Risk, Compliance, Data Science
• Every asset can be assigned to one domain -- set via the sidebar, replaces the existing assignment
• Domains are set up by admins, but visible and useful for everyone
• What domains enable:
• Navigation: browse the "Finance domain" to see all Finance assets in one place
• Governance: policies and Data Products can be scoped per domain
• Discovery: filter any search by domain to narrow results instantly

• Every asset in DataHub can be described with these four concepts working together
• Example: the FACT_REVENUE_DAILY Snowflake table
• Entity: PROD.FINANCE.FACT_REVENUE_DAILY (a Dataset entity)
• URN: urn:li:dataset:(urn:li:dataPlatform:snowflake,PROD.FINANCE.FACT_REVENUE_DAILY,PROD)
• Aspects: SchemaMetadata -- 47 columns, types; Ownership -- owned by finance-data-team group
• Domain: Finance (the business domain this dataset belongs to)

• Your task (5 minutes):
• Navigate to the live DataHub instance shared in the session setup
• Search for any dataset in the catalog
• Open the entity page and locate:
• The entity's URN (click "Copy URN" or read it from the URL)
• The domain this entity belongs to
• At least two aspects visible on the page (e.g., the Columns tab = SchemaMetadata, the Owners section in the sidebar = Ownership aspect)

• Rule: an asset belongs to the domain of the team responsible for its quality, governance, and lifecycle
• Not: the domain of every team that queries it
• Example: FACT_REVENUE_DAILY is owned and governed by the Finance data team
• It is also consumed by Marketing and Executive Reporting
• Correct domain: Finance (governance determines domain, not consumption)
• Consumption relationships are tracked separately via lineage and subscriptions

• I have an Airflow DAG called daily_revenue_pipeline. What entity type is it in DataHub, and what would its URN look like?
• A data engineer says "the entity is missing its owner." What is wrong with this statement, and what information is missing?
• Your company has a dataset that is used by the Finance team but owned and governed by the Data Engineering team. Which domain should it be assigned to?

• Entity: a specific trackable thing in DataHub (Dataset, Dashboard, Pipeline, User, etc.)
• Aspect: a named bundle of properties on an entity (Schema, Ownership, Tags, Profile...)
• URN: the permanent, unique identifier for every entity -- use it in scripts and API calls
• Domain: the business grouping that places an asset in the org chart and scopes governance
• Key rule to remember: entity type vs. entity (specific instance) -- always be specific