Monitor Data Boost usage

This page describes how to use Cloud Monitoring and Spanner audit logs to monitor and analyze Spanner Data Boost usage.

Cloud Monitoring provides the ability to track total usage over time and to break down that usage by user. Spanner audit logs allow more flexible usage analysis, including providing metrics by user or by BigQuery job ID.

Use Cloud Monitoring to track Data Boost usage

Follow these steps to track overall Data Boost usage:

  1. In the Google Cloud console, go to Monitoring.
    Go to Monitoring
  2. If Metrics Explorer is shown in the navigation menu, select it. Otherwise, click Resources, and then select Metrics Explorer.
  3. At the top of the page, select a time interval.
  4. In the Metric dropdown list, in the Filter by resource or metric name field, enter spanner and press Enter to narrow the search.

  5. In the Metric list, select Cloud Spanner Instance > Instance > Processing Unit Second, and then click Apply.

    This creates a line chart of aggregate Data Boost usage across all Spanner instances.

  6. To view usage for a particular instance, follow these steps:

    1. Click + ADD FILTER to create a filter.
    2. In the Label 1 dropdown list, select instance_id.
    3. In the Value 1 text field, enter or select the ID of the instance to display usage for.

  7. To view a breakdown of usage by all instances, follow these steps:

    1. Clear any filters by clicking the trash can icon next to the Filters fields.
    2. Under Group by, in the Labels dropdown list, select instance_id, and then select Sum in the Grouping function list.

  8. To break down usage by principal, under Group by, select credential_id in the Labels list and Sum in the Grouping function list.

The following example shows processing unit seconds by principal.

PU seconds by user

Use audit logs to analyze Data Boost usage

Spanner audit logs allow more flexible analysis of Data Boost usage. In addition to the ability to break down usage over time by instance or principal as with Cloud Monitoring, Spanner audit logs, if enabled and available, allow breaking down usage over time by database or BigQuery job ID.

Enabling audit logs can incur extra charges. For information about Logging pricing, see Google Cloud Observability pricing: Cloud Logging.

Enable Data Boost usage audit logs

You must enable data access audit logs for Spanner before you can collect usage data for Data Boost. To do so, follow these steps:

  1. Follow the instructions in Configure Data Access audit logs with the Google Cloud console.
  2. Enable the Data Read log type for the Spanner API service.
  3. To obtain Data Boost usage by BigQuery job SQL text, ensure that that the audit logs for BigQuery are enabled as well.

View usage by principal

To query the audit logs to view Data Boost usage by user, follow these steps:

  1. Go to the Logs Explorer in the Google Cloud console.

    Go to Logs Explorer

  2. In the navigation menu, click Log Analytics.

  3. To show usage by user and database over the past 7 days, run the following query. To change the timespan for which usage is shown, modify the timestamp expression in the WHERE clause.

    SELECT
  SUM(CAST(JSON_VALUE(labels.data_boost_usage) AS INT64)) AS usage,
  REGEXP_EXTRACT(
    proto_payload.audit_log.resource_name,
    'projects/[^/]+/instances/[^/]+/databases/[^/]+') AS database,
  proto_payload.audit_log.authentication_info.principal_email AS principal_email
FROM `PROJECT_NAME.global._Default._AllLogs`
WHERE
  timestamp > TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL 7 DAY)
  AND resource.type = 'spanner_instance' AND operation.last IS NULL
  AND JSON_VALUE(labels.data_boost_usage) != ''
GROUP BY database, principal_email;

    Replace PROJECT_NAME with your project name.

The following example shows usage in processing units for 4 principals.

Data Boost usage by user from audit logs

View usage by BigQuery job ID

To query the audit logs to view Data Boost usage broken down by database, user, and BigQuery job ID, follow these steps:

  1. Go to the Logs Explorer in the Google Cloud console.

    Go to Logs Explorer

  2. In the navigation menu, click Log Analytics.

  3. Run the following query:

    SELECT
  SUM(CAST(JSON_VALUE(labels.data_boost_usage) AS INT64)) AS usage,
  REGEXP_EXTRACT(
    proto_payload.audit_log.resource_name,
    'projects/[^/]+/instances/[^/]+/databases/[^/]+') AS database,
  proto_payload.audit_log.authentication_info.principal_email AS principal_email,
  IFNULL(JSON_VALUE(labels.data_boost_workload_id), 'not from BQ') AS job_id
FROM `PROJECT_NAME.global._Default._AllLogs`
WHERE
  timestamp > TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL 7 DAY)
  AND resource.type = 'spanner_instance' AND operation.last IS NULL
  AND JSON_VALUE(labels.data_boost_usage) != ''
GROUP BY database, principal_email, job_id;

    Replace PROJECT_NAME with your project name.

The following example shows usage by BigQuery job ID.

Data Boost usage by job ID

View usage by BigQuery SQL text

To view Data Boost usage for multiple BigQuery jobs aggregated by the SQL text of those jobs, follow these steps:

  1. Go to the Logs Explorer in the Google Cloud console.

    Go to Logs Explorer

  2. In the navigation menu, click Log Analytics.

  3. Run the following query:

    SELECT
  SUM(
    CAST(
      JSON_VALUE(db.labels.data_boost_usage)
      AS INT64)) AS usage,
  JSON_VALUE(
    bq.proto_payload.audit_log.metadata.jobInsertion.job.jobConfig.queryConfig.query)
    AS bq_query
FROM
  `PROJECT_NAME.global._Default._AllLogs` db,
  `PROJECT_NAME.global._Default._AllLogs` bq
WHERE
  db.timestamp > TIMESTAMP_SUB(
    CURRENT_TIMESTAMP(), INTERVAL 7 DAY)
  AND db.resource.type = 'spanner_instance'
  AND JSON_VALUE(db.labels.data_boost_usage) != ''
  AND db.operation.last IS NULL
  AND bq.timestamp > TIMESTAMP_SUB(
    CURRENT_TIMESTAMP(), INTERVAL 7 DAY)
  AND bq.proto_payload.audit_log.method_name
    = 'google.cloud.bigquery.v2.JobService.InsertJob'
  AND bq.resource.type = 'bigquery_project'
  AND JSON_VALUE(
    bq.proto_payload.audit_log.metadata.jobInsertion.job.jobConfig.queryConfig.query)
    IS NOT NULL
  AND JSON_VALUE(db.labels.data_boost_workload_id)
    = REGEXP_EXTRACT(bq.proto_payload.audit_log.resource_name, '[^/]*$')
GROUP BY bq_query
ORDER BY usage DESC

    Replace PROJECT_NAME with your project name.

The following example shows Data Boost usage by SQL text.

Data Boost usage by SQL text

Create a Data Boost alert

To create an alert that is issued when Data Boost usage exceeds a predefined threshold, see Set an alert for Data Boost usage.

What's next