About Observability

You can use the Observability pages in the web interface to monitor your Fastly CDN and Compute services.

Before you begin

About the Observability pages

When you click Observability in the navigation sidebar, the Account Summary page appears by default. The Account Summary page provides a high-level overview of all your services.

To access additional Observability pages, go to Observability in the navigation sidebar and select one of the following options:

• Services menu: provides access to the Service Overview page, which displays top metrics for a Fastly service, and the All Services page, which displays aggregated metrics for all of your services.
• Domains page: provides real-time and historic visibility into detailed response data for traffic from your domains and subdomains to Fastly.
• Origins page: provides real-time and historic visibility into detailed response data for traffic from your origin servers to Fastly.
• Logs menu: provides access to the Log tailing page, where you can view custom log messages from your Compute application.
• Alerts menu: provides access to the Overview, Definitions, and Integrations pages, where you can view current and historical alert feeds and configure alert definitions and integrations.
• Custom Dashboards page: where you can view and manage custom dashboards that surface relevant metrics for a Fastly service.

The exact graphs displayed on the Observability pages depend on your access permissions and the type of service you create, as well as other products and features you may have purchased.

The data on the Observability pages may also appear grayed out or blank to some users, with no information displayed in the controls, when a service hasn't yet received enough requests for Fastly to display meaningful information about it.

About the controls on the Observability pages

On the Account Summary, Service Overview, All Services, Origins, Domains, and Custom Dashboards pages, a series of controls appear to the right of and below the dashboard name. These controls allow you to refine the displayed metrics on the dashboard. Specifically:

• the Alerts link, which opens the Alerts firing window. From this window, you can view firing alerts.
• the Dashboard Name menu (which only appears on dashboards for specific services, not the Account Summary dashboard) allows you to select and view a different dashboard.
• the Options menu allows you to copy a link to the dashboard, export the data from the dashboard to a CSV file, and display event markers on the charts. Event markers are vertical lines that signify when a new version of the service was activated.
• the Service configuration link opens the service configuration page for the service selected in the dashboard.
• the Create dashboard button allows you to create a custom dashboard.
• the Service menu allows you to specify the service that the metrics are drawn from. From the menu, you can search for a service by ID or name.
• the Region menu allows you to limit the displayed metrics on any dashboard to a specific region around the world.
• the Data Resolution menu allows you to select how the data is represented for the defined time range.
• the Time Range menu allows you to change the timeframe over which metrics will be displayed and to change the time zone from Coordinated Universal Time (UTC) to the time zone identified by your browser. By default, the system displays 1 days worth of static data. The selected timeframe and time zone persist as you navigate between dashboards.
• the Live button (which only appears on dashboards for specific services, not the Account Summary dashboard) allows you to change displayed metrics to live data that changes in real time. When viewing the live data stream, the time axis will scroll even if there is no data. You can click Pause to pause the live data stream.

You can also learn more information about a chart by:

• hovering over any part of a graph. This displays a timestamp indicator that updates as you move the cursor along with a service version indicator.
• using the summary menu at the bottom of a chart to view the total, maximum, minimum, mean, latest, and 95th percentile values of a graph. By default, percentile charts display the mean value while other charts display the total. The summary menu for percentile charts does not include the total metric.

In addition, on dashboards that aren't custom, you can click expand in the upper right corner of a chart to access the Metric Details page, which contains detailed information about the metric being tracked. Specifically, the Metric Details page displays:

• the expanded chart.
• a summary table that automatically updates the total, maximum, minimum, mean, and 95th percentile values for the metric based on the timestamp that you're hovering over on the expanded chart.
• a table that displays the total for the metric over time.

About the Account Summary page

The Account Summary page provides aggregate metrics for your account for the last two weeks:

• Requests: the total number of CDN and Compute requests aggregated across your account.
• Bandwidth: the total amount of bandwidth served across all services in your account.
• Error ratio: the average ratio of error responses (4xx and 5xx) to the total number of requests across your account.

Links to your latest alerts and custom dashboards are listed at the bottom of the page. Alerts and custom dashboards are disabled by default.

About data latency

When working with dashboards and the metrics they display, keep the following in mind about data:

• We do not report data for time periods that have not yet ended.
• Data with day resolution is bucketed based on UTC days and each day's data becomes available around 2am the following day.
• Data with hour resolution buckets becomes available approximately 15 minutes after the end of each hour.
• Data with minute resolution buckets usually becomes available two minutes after the end of the minute, but can take up to 15 minutes to appear.

If your use case requires data closer to real-time, consider using the generally available real-time API instead.

What's next