Google Cloud Platform observability

Google Cloud Platform (GCP) observability allows you to monitor GCP metrics and view GCP logs.

GCP Metrics

Grafana Alloy sends metrics from Cloud Monitoring to Grafana Cloud. Alloy uses its embedded stackdriver-exporter to export metrics from Cloud Monitoring and then send them to Grafana Cloud.

To get GCP Metrics running, refer to Configure GCP Metrics.

Preconfigured dashboards

Preconfigured dashboards are out-of-the-box visualizations available in Grafana Cloud for popular Google Cloud services.

Cloud Provider Observability provides preconfigured dashboards for the following GCP services. Click the link to see details for the metrics provided for these services.

AlloyDB for PostgreSQL: AlloyDB
Bigtable: Cloud Bigtable
Cloud Run: Cloud Run
Cloud SQL: Cloud SQL
Compute Engine: Compute Engine
Cloud Load Balancing: Load Balancing
Pub/Sub: Pub/Sub
Cloud Storage: Storage (GCS)
Virtual Private Cloud: VPC networks

Refine dashboard data

You can use the filters on any dashboard to refine your data. Filters are appropriate for each dashboard, and may include:

Data source
Job
Project ID
Instance
Database ID
Bucket name
Country
Backend target
Subscription ID

Filtering a dashboard to refine data by subscription ID

Additionally, use the time range selector to change time period of your data.

Changing the time range selector for a dashboard to the last two days

Go to a GCP dashboard

To see any metrics dashboard, configure GCP metrics.
In the main menu, click GCP to open the Services tab.
List of available services on Services tab
Locate the specific service in the list, and click the dashboard in the View your data column of the table.

View predictions

For the preconfigured dashboards that include drilldown information for specific instances, Cloud Provider Observability includes machine learning predictions. Predictions can help you ensure resources are available during spikes in usage, as well as help you decrease the amount of unused resources due to over provisioning. To use prediction tools, first enable LLM features for your Grafana instance.

You can view the prediction model for various metrics by clicking the Predict button in the top right corner of the panel.

Screenshot of a panel with the Predict button and time range filter highlighted

Use the time range filter to adjust the time range to show more advanced predictions. The time range you select must be at least two hours to use the prediction tool.

For more information on the terminology included and how machine learning works in the prediction graph, refer to the Query Metrics page in the AI and machine learning documentation.

GCP Logs

Grafana Alloy sends logs from Cloud Monitoring to Grafana Cloud. Logs are sent from Cloud Monitoring through a log sink to a Pub/Sub topic. The GCP SDK is used for pulling and receiving messages from Pub/Sub. The service account enables Alloy to read the logs from the Pub/Sub subscription. Alloy then sends the logs to Grafana Cloud.