Service health

Cloud Observability offers a way to quickly see how all your services and their operations are performing in one place - the Service Directory view.

You can also use our pre-built service dashboards or the Service health panel to view service health.

From here, you can:

• See all your services in one place
  • Search for a service
  • Filter to view by language
  • Mark “favorite” services
• See performance over time and the top performance changes for your Key Operations
  • View service health after deployments
  • Compare latency, error, and infrastructure regression data against baseline data
  • View and sort latency, operation rate, and error rate and dive into root cause analysis
• View all operations (server and client) on a service
  • View current performance of individual operations or performance at a specific percentile or performance change over a period of time
  • Filter by service or client operations
• View Streams (retained span queries).

Access the service directory view

When you first open Cloud Observability, you’re taken to the Service Directory. You can also access it from the navigation bar.

Find services

Your services are listed in alphabetical order. To make finding services easier, you can “favorite” a service so it always appears at the top of the list.

To find a service:

• Use the Search box to search for services by name. As you type, Cloud Observability filters the list of services that match your entry.
• Filter by language. All services show the language of the associated instrumentation. Use the language dropdown at the top of the list to filter by a language.

To favorite a service:

1. Find and select the service to favorite.
2. Click the star next to the service’s name. The service now appears at the top of the list.

View service health, see top changes, and compare operation performance

We will be introducing new workflows to replace the RCA view. As a result, it will soon no longer be supported. Instead, use notebooks for your investigation where you can run ad-hoc queries, view data over a longer time period, and run Cloud Observability’s correlation feature.

The Service Health view on the Deployments tab shows you the latency, error rate, and operation rate of your Key Operations (operations whose performance is strategic to the health of your system) on the selected service.

Key Operations are displayed in order of magnitude of change in performance (you can change the order). You can quickly see the top performance changes and compare performance during latency or error regression spikes with normal performance.

About key operations

By default, Cloud Observability dynamically determines key operations as the ingress or root operations that have the highest rate for that service. You can also manually select key operations so that they always display in this view.

By default, each service has 30 key operations. So if you manually select 10 key operations, Cloud Observability dynamically chooses 20 of the remaining highest rate ingress operations to display as well.

You can change the default number of key operations. Contact your Customer Success representative for more information.

To manually set key operations:

1. On the Deployments tab of the Service Directory, click Edit next to Key Operations.

2. In the dialog, select the operations you always want to be key operations, up to 30 per service. If you select less than 30, Cloud Observability dynamically determines the remaining key operations for you.

Change the sparkline charts display

By default, the operations are sorted by the amount of detected change (largest to smallest). Use the dropdown to change the sort.

Also by default, only the latency percentile with the largest amount of change displays. You can change the sparkline charts to show more percentiles using the More ( ⋮ ) icon.

You can search for an operation using the Search field.

Change the reporting time period

By default, the data shown is from the last 60 minutes. You can customize that time period using the time picker. Use the < > controls to move backwards and forwards through time. You can view data from your retention window (default is three days).

You can also zoom in on a time period by clicking and dragging over the time period you want a closer look at. The charts redraw to report on just the period you selected.

View top changes in performance

Cloud Observability displays sparkline charts for Key Operations on the service. They show recent performance for Service Level Indicators (SLIs): latency, error rate, and operation rate. Shaded yellow bars to the left of the chart indicate the magnitude of the change.

The operations are initially sorted by highest change during the visible time period (you can change that time period) and you can also change the sorting order).

If there’s a deployment marker visible, that change is measured as the difference between the selected version and all other versions. If there is no marker, the change is measured as the difference between the first and second half of the time period shown in the chart.

More About How Cloud Observability Measures Change

Cloud Observability measures two aspects of change: size and continuity. A full bar indicates that a large, sustained change has happened. Smaller bars indicate either a smaller change or one that did not last for the full time period. Only changes that are relative (i.e. a change of 10ms to 500ms is ranked higher than one of 1s to 2s) are considered.

The yellow bar means that an SLI had an objectively large change, regardless of service or operation. Cloud Observability’s algorithm runs on each SLI independently. For example, when the bar displays for an operation’s latency, that means latency has changed – not that its change was greater compared to the other SLIs.

When determining change, Cloud Observability compares the SLI of the baseline SLI time series to the comparison SLI time series. Those time periods are determined using the data currently visible in the charts.

You can change the amount of time displayed using the time period dropdown at the top right of the page.

The baseline and comparison time periods are determined as follows:

If there is one or more deployment markers visible:

• For latency and error rates, Cloud Observability compares the performance after the selected version to performance in all other versions.
• For operation rate, it compares the rate before the deployment marker to after the deployment marker.

If there are no deployment markers visible:
Cloud Observability compares the performance of the first half of the time period to the second half.

Expandable end

Select an operation’s sparkline charts to view larger charts for the latency, error rate, and operation rate. You use these larger charts to start your investigation.

View associated spans

Span samples for the selected operation are shown in a table below the charts (click View spans). Click a span’s row to view the span in it’s trace.

Monitor deployments

The Service Health view on the Deployments tab provides the ability to see how deployments specifically affect performance of your services. When you implement an attribute to display versions of your service, a deployment marker displays at the time the deployment occurred on all charts in Cloud Observability.

We will be introducing new workflows to replace the RCA view. As a result, it will soon no longer be supported. Instead, use notebooks for your investigation where you can run ad-hoc queries, view data over a longer time period, and run Cloud Observability’s correlation feature.

These markers allow you to quickly correlate deployment with a possible regression.

When you have multiple versions in a time window, you can view the performance of each deployed version. For example, in this image of the Service Health view, multiple versions have been deployed. Hover over the chart to see the percentage of traffic in each version.

Learn more about how to use this view to monitor the health of your deployments.

Compare performance from different time ranges

We will be introducing new workflows to replace the RCA view. As a result, it will soon no longer be supported. Instead, use notebooks for your investigation where you can run ad-hoc queries, view data over a longer time period, and run Cloud Observability’s correlation feature.

When you spot a latency or error rate regression, you can start an investigation by clicking the corresponding time series chart during the regression. You can choose to compare performance from before the previous deploy, an hour ago, a day ago, or select a custom baseline.

Choose a time in the middle of the regression to avoid collecting data previous to the spike.

Latency root cause analysis

We will be introducing new workflows to replace the RCA view. As a result, it will soon no longer be supported. Instead, use notebooks for your investigation where you can run ad-hoc queries, view data over a longer time period, and run Cloud Observability’s correlation feature.

This view provides the following tools to help you with root cause analysis for latency:

Learn how to use these tools here.

Error rate root cause analysis

We will be introducing new workflows to replace the RCA view. As a result, it will soon no longer be supported. Instead, use notebooks for your investigation where you can run ad-hoc queries, view data over a longer time period, and run Cloud Observability’s correlation feature.

Cloud Observability offers these tools for analyzing spikes in error rate:

Learn how to use these tools here.

Add a time series to a notebook

You can add any of the time series charts to a notebook for when, during an investigation, you want to be able to run ad hoc queries, take notes, and save your analysis for use in postmortems or runbooks. Notebooks let you view logs, metrics, and traces from different places in Cloud Observability together, in one place.

To add to a notebook, click Add to notebook and search to choose an existing notebook or create a new notebook.

When you add to a notebook, a panel is created using the same query. You can see the latency for multiple percentiles and view exemplar traces. The annotation is a link back to the original, so you can quickly return to the origin of your investigation.

Learn more about notebooks.

Create a dependency map

To view the relationships of the selected service and operation to upstream and downstream services and operations, click Create dependency map and add it to a notebook.

View operation performance

The Operations tab on the Service Directory view shows the selected service’s operations currently reporting to Cloud Observability in alphabetical order, along with performance metrics aggregated over the selected time period.

The table provides several useful performance metrics for each operation:

• Latency Change: Change in latency between now and the time period set using the Change Since dropdown.
• Latency: How long the operation took to complete for a given percentile, set using the Percentile dropdown.
• Error Change: The percentage change in error rate for the time period set using the Change Since dropdown.
• Errors: The percentage of operation instances that contain an error.
• Rate Change: The percentile change of rate in the time period set using the Change Since dropdown.
• Rate: The number of times the operation occurred per second.
• View stream: Add a chart to a notebook or dashboard using the query for the operation. Click the row to view the associated Stream.
• Create stream: Create a Stream for the operation. Creating a Stream for an operation allows you to retain the data for the associated query for longer than your retention window.

Search and filter operations

• Use the Search box to search for operations by name. As you type, Cloud Observability filters the list of operations that match your entry.
• View only server or client operations by clicking the respective tab. Server operations are the first operations handling external requests from outside that service (i.e. API HTTP GET etc.). Client operations are those that call out to external services.

To see if other services are affecting an operation, view the operation in a notebook or dashboard and use the dependency map to view upstream and downstream services and their performance.

View streams for an operation or service

Streams are retained span queries that continuously collect latency, error rate and operation rate data. By default, data from span queries are persisted for three days. When you save a query as a Stream, the data is collected and persisted for a longer period of time.

To view all Streams for a service, click the Streams tab. The number on the tab tells you how many Streams exist for this service.

Create a Stream from the Operations tab by clicking Create Stream for an operation.

Add a Stream’s query to a notebook or dashboard

You can add charts that show the Stream’s performance to either a notebook or a dashboard. When you add a Stream, three charts are created: one for latency, one for error rate, and one for operation rate.

Add an Stream’s query to a notebook for when, during an investigation, you want to be able to run ad hoc queries, take notes, and save your analysis for use in postmortems or runbooks. Notebooks allow you to view metric and trace data from different places in Cloud Observability together, in one place.

Add the query to a dashboard when you want to monitor the performance over a period of time.

View a service’s dashboards

Click the Dashboards tab to view dashboards that include charts or a Stream for this service. The number on the tab tells you how many dashboards exist for this service.

Only dashboards that have charts that contain a filter for the service are shown.

Click a dashboard to view it.

Read Create and manange dashboards to learn more.

View and improve your instrumentation quality

The data you can view and use in Cloud Observability depends on the quality of your tracing instrumentation. The better and more comprehensive your instrumentation is, the better Cloud Observability can collect and analyze your data to provide highly actionable information.

Cloud Observability analyzes the instrumentation on your services and determines how you can improve it to make your Cloud Observability experience even better. It can determine whether you instrumentation:

• Crosses services to create full traces
• Includes interior spans to help find the critical path
• Contains attributes to help find correlated areas of latency. If there are attributes that you’d like all services to report to Cloud Observability (like a customer ID or Kubernetes region), you can register the corresponding attributes and Cloud Observability will check for those when determining the IQ score.
• Uses attributes for deployments to help monitor regressions
• Contains hostname attributes to help find performance issues in different environments.

Click the Instrumentation Quality tab to learn how well your instrumentation measures up. The number on the tab gives your score (based on 100%).

Learn more about what your score means and how to fix it.

See also

Create and manage dashboards

Notebooks

Monitor deployments

Updated Apr 1, 2024