Zdieľať cez

Monitor usage with system tables

  • Článok

This article explains the concept of system tables in Azure Databricks and highlights resources you can use to get the most out of your system tables data.

What are system tables?

System tables are an Azure Databricks-hosted analytical store of your account’s operational data found in the system catalog. System tables can be used for historical observability across your account.

Note

For documentation on system.information_schema, see Information schema.

Requirements

  • To access system tables, your workspace must be enabled for Unity Catalog. For more information, see Enable system table schemas.

  • System tables are not available in the following regions:

    • Azure China regions
    • Azure Government regions
    • West India
    • Switzerland West

Which system tables are available?

Currently, Azure Databricks hosts the following system tables:

Table Description Location Supports streaming Free retention period Includes global or regional data
Audit logs (Public Preview) Includes records for all audit events from workspaces in your region. For a list of available audit events, see Diagnostic log reference. system.access.audit Yes 365 days Regional for workspace-level events. Global for account-level events.
Table lineage (Public Preview) Includes a record for each read or write event on a Unity Catalog table or path. system.access.table_lineage Yes 365 days Regional
Column lineage (Public Preview) Includes a record for each read or write event on a Unity Catalog column (but does not include events that do not have a source). system.access.column_lineage Yes 365 days Regional
Billable usage Includes records for all billable usage across your account. Each usage record is an hourly aggregate of a resource’s billable usage. system.billing.usage Yes 365 days Global
Pricing A historical log of SKU pricing. A record gets added each time there is a change to a SKU price. system.billing.list_prices No Unlimited Global
Clusters (Public Preview) A slow-changing dimension table that contains the full history of compute configurations over time for any cluster. system.compute.clusters Yes N/A Regional
Node timeline (Public Preview) Captures the utilization metrics of your all-purpose and jobs compute resources. system.compute.node_timeline Yes 30 days Regional
Node types (Public Preview) Captures the currently available node types with their basic hardware information. system.compute.node_types No N/A Regional
SQL warehouse events (Public Preview) Captures events related to SQL warehouses. For example, starting, stopping, running, scaling up and down. system.compute.warehouse_events Yes 365 days Regional
Jobs (Public Preview) Tracks all jobs created in the account. system.lakeflow.jobs Yes 365 days Regional
Job tasks (Public Preview) Tracks all job tasks that run in the account. system.lakeflow.job_tasks Yes 365 days Regional
Job run timeline (Public Preview) Tracks the start and end times of job runs. system.lakeflow.

job_run_timeline		 Yes 365 days Regional
Job task timeline (Public Preview) Tracks the start and end times and compute resources used for job task runs. system.lakeflow.

job_task_run_timeline		 Yes 365 days Regional
Marketplace funnel events (Public Preview) Includes consumer impression and funnel data for your listings. system.marketplace.listing_

funnel_events		 Yes 365 days Regional
Marketplace listing access (Public Preview) Includes consumer info for completed request data or get data events on your listings. system.marketplace.listing_

access_events		 Yes 365 days Regional
Predictive optimization (Public Preview) Tracks the operation history of the predictive optimization feature. system.storage.predictive_

optimization_operations_history		 No 180 days Regional
Databricks Assistant events (Public Preview) Tracks user messages sent to the Databricks Assistant. system.access.assistant_events No 365 days Regional
Query history (Public Preview) Captures records for all queries run on SQL warehouses. system.query.history Yes 90 days Regional
Clean room events (Public Preview) Captures events related to clean rooms. system.access.clean_room_events Yes 365 days Regional
Model serving endpoint usage (Public Preview) Captures token counts for each request to a model serving endpoint and its responses. system.serving.endpoint_usage Yes 90 days Regional
Model serving endpoint data (Public Preview) A slow changing dimension table that stores metadata for each served external model in a model serving endpoint. system.serving.served_entities Yes 365 days Regional

The billable usage and pricing tables are free to use. Tables in Public Preview are also free to use during the preview but could incur a charge in the future.

Note

You may see other system tables in your account besides the ones listed above. Those tables are in Private Preview currently and are empty by default. If you are interested in using any of these tables, please reach out to your Databricks account team.

Enable system table schemas

Since system tables are governed by Unity Catalog, you need to have at least one Unity Catalog-enabled workspace in your account to enable and access system tables. System tables include data from all workspaces in your account but they can only be accessed from a Unity Catalog-enabled workspace.

System tables are enabled at the schema level. If you enable a system schema, you enable all the tables within that schema. When new schemas are released, an account admin needs to manually enable the schema.

System tables must be enabled by an account admin. You can enable system tables using the SystemSchemas API.

Note

The billing schema is enabled by default. Other schemas must be enabled manually.

List available system schemas

Use the following curl command to list available system schemas:

curl -v -X GET -H "Authorization: Bearer <PAT Token>" "https://adb-<xxx>.azuredatabricks.net/api/2.0/unity-catalog/metastores/<metastore-id>/systemschemas"

The following is an example output of the GET command:

{"schemas":[{"schema":"access","state":"<AVAILABLE OR EnableCompleted>"},{"schema":"billing","state":"<AVAILABLE OR EnableCompleted>"},{"schema":"information_schema","state":"<AVAILABLE OR EnableCompleted>"}]}

state: AVAILABLE: The system schema is available but has not yet been enabled.

state: EnableCompleted: You have enabled the system schema and it is visible in Catalog Explorer.

Enable a system schema

Use the following curl command to enable a system schema:

curl -v -X PUT -H "Authorization: Bearer <PAT Token>" "https://adb-<xxx>.azuredatabricks.net/api/2.0/unity-catalog/metastores/<metastore-id>/systemschemas/<SCHEMA_NAME>"

If the system schema is enabled successfully, result code 200 is returned.

If you attempt to re-enable a system schema, the following is returned: "error_code":"SCHEMA_ALREADY_EXISTS","message":"Schema <schema-name> already exists".

Disable a system schema

Use the following curl command to disable a system schema:

curl -v -X DELETE -H "Authorization: Bearer <PAT Token>" "https://adb-<xxx>.azuredatabricks.net/api/2.0/unity-catalog/metastores/<metastore-id>/systemschemas/<SCHEMA_NAME>"

Grant access to system tables

Access to system tables is governed by Unity Catalog. No user has access to these system schemas by default. To grant access, a user that is both a metastore admin and an account admin must grant USE and SELECT permissions on the system schemas. See Manage privileges in Unity Catalog.

System tables are read-only and cannot be modified.

Note

If your account was created after November 9, 2023, you might not have a metastore admin by default. For more information, see Set up and manage Unity Catalog.

Do system tables contain data for all workspaces in your account?

System tables contain operational data for all workspaces in your account deployed within the same cloud region. Billing system tables contain account-wide data.

Even though system tables can only be accessed through a Unity Catalog workspace, the tables also include operational data for the non-Unity Catalog workspaces in your account.

Where is system table data stored?

Your account’s system table data is stored in a Azure Databricks-hosted storage account located in the same region as your metastore. The data is securely shared with you using Delta Sharing.

Each table has a free data retention period. For information on extending the retention period, contact your Azure Databricks account team.

Where are system tables located in Catalog Explorer?

The system tables in your account are located in a catalog called system, which is included in every Unity Catalog metastore. In the system catalog you’ll see schemas such as access and billing that contain the system tables.

Considerations for streaming system tables

Azure Databricks uses Delta Sharing to share system table data with customers. Be aware of the following considerations when streaming with Delta Sharing:

  • If you are using streaming with system tables, set the skipChangeCommits option to true. This ensures the streaming job is not disrupted from deletes in the system tables. See Ignore updates and deletes.
  • Trigger.AvailableNow is not supported with Delta Sharing streaming. It will be converted to Trigger.Once.

If you use a trigger in your streaming job and find it isn’t catching up to the latest system table version, Databricks recommends increasing the scheduled frequency of the job.

Read incremental changes from streaming system tables

spark.readStream.option("skipChangeCommits", "true").table("system.billing.usage")

Known issues

  • Currently no support for real-time monitoring. Data is updated throughout the day. If you don’t see a log for a recent event, check back later.

  • To enable system tables, you might need to grant network access to the system tables Blob storage endpoint. To view a list of every region’s system tables’ storage endpoint, see Storage endpoint IP addresses.

  • The system schemas system.operational_data and system.lineage are deprecated and will contain empty tables.

  • The __internal_logging system table schema is used to support payload logging using inference tables. This schema is visible to account admins, but it cannot be enabled and should not be used for customer workflows.