Structured properties are a structured, named set of properties that can be attached to logical entities like Datasets, DataJobs, etc.
Structured properties have values that are types. Conceptually, they are like “field definitions”.
Learn more about structured properties in the [Structured Properties Feature Guide](../../../docs/features/feature-guides/properties.md).
### Goal Of This Guide
This guide will show you how to execute the following actions with structured properties.
- Create structured properties
- Read structured properties
- Delete structured properties (soft delete)
- Add structured properties to a dataset
- Patch structured properties (add / remove / update a single property)
## Prerequisites
For this tutorial, you need to deploy DataHub Quickstart and ingest sample data.
For detailed information, please refer to [Datahub Quickstart Guide](/docs/quickstart.md).
Additionally, you need to have the following tools installed according to the method you choose to interact with DataHub:
<Tabs>
<TabItemvalue="CLI"label="CLI"default>
Install the relevant CLI version. Forms are available as of CLI version `0.13.1`. The corresponding SaaS release version is `v0.2.16.5`
Connect to your instance via [init](https://datahubproject.io/docs/cli/#init):
- Run `datahub init` to update the instance you want to load into.
- Set the server to your sandbox instance, `https://{your-instance-address}/gms`.
- Set the token to your access token.
</TabItem>
<TabItemvalue="OpenAPI"label="OpenAPI">
Requirements for OpenAPI are:
* curl
* jq
</TabItem>
</Tabs>
## Create Structured Properties
The following code will create a structured property `io.acryl.privacy.retentionTime`.
<Tabs>
<TabItemvalue="CLI"label="CLI"default>
Create a yaml file representing the properties you’d like to load.
For example, below file represents a property `io.acryl.privacy.retentionTime`. You can see the full example [here](https://github.com/datahub-project/datahub/blob/example-yaml-sp/metadata-ingestion/examples/structured_properties/struct_props.yaml).
```yaml
- id: io.acryl.privacy.retentionTime
# - urn: urn:li:structuredProperty:io.acryl.privacy.retentionTime # optional if id is provided
qualified_name: io.acryl.privacy.retentionTime # required if urn is provided
type: number
cardinality: MULTIPLE
display_name: Retention Time
entity_types:
- dataset # or urn:li:entityType:datahub.dataset
- dataFlow
description: "Retention Time is used to figure out how long to retain records in a dataset"
allowed_values:
- value: 30
description: 30 days, usually reserved for datasets that are ephemeral and contain pii
- value: 90
description: Use this for datasets that drive monthly reporting but contain pii
- value: 365
description: Use this for non-sensitive data that can be retained for longer
```
Use the CLI to create your properties:
```commandline
datahub properties upsert -f {properties_yaml}
```
If successful, you should see `Created structured property urn:li:structuredProperty:...`
"description": "Retention Time is used to figure out how long to retain records in a dataset",
"entityTypes": [
"urn:li:entityType:datahub.dataset",
"urn:li:entityType:datahub.dataFlow"
],
"cardinality": "MULTIPLE"
}
}
```
</TabItem>
</Tabs>
## Set Structured Property To a Dataset
This action will set/replace all structured properties on the entity. See PATCH operations to add/remove a single property.
<Tabs>
<TabItemvalue="CLI"label="CLI"default>
You can set structured properties to a dataset by creating a dataset yaml file with structured properties. For example, below is a dataset yaml file with structured properties in both the field and dataset level.
Please refer to the [full example here.](https://github.com/datahub-project/datahub/blob/example-yaml-sp/metadata-ingestion/examples/structured_properties/datasets.yaml)
Following command will set structured properties `retentionTime` as `60.0` to a dataset `urn:li:dataset:(urn:li:dataPlatform:hive,SampleHiveDataset,PROD)`.
Please note that the structured property and the dataset must exist before executing this command. (You can create sample datasets using the `datahub docker ingest-sample-data`)
There are two types of deletion present in DataHub: hard and soft delete. As of the current release only the soft delete is supported for Structured Properties.
:::note SOFT DELETE
A soft deleted Structured Property does not remove any underlying data on the Structured Property entity or the Structured Property's values written to other entities. The soft delete is 100% reversible with zero data loss. When a Structured Property is soft deleted, a few operations are not available.
Structured Property Soft Delete Effects:
- Entities with a soft deleted Structured Property value will not return the soft deleted properties
- Updates to a soft deleted Structured Property's definition are denied
- Adding a soft deleted Structured Property's value to an entity is denied
- Search filters using a soft deleted Structured Property will be denied
If you want to **remove the soft delete**, you can do so by either hard deleting the status aspect or changing the removed boolean to `false` like below.