After you've sent us your OpenTelemetry data and opened your service (entity) in the UI, the Summary page offers an overview of your service's health. Here you can see:
- The golden signals for your service: response time, throughput, and error rate
- Entities this service depends on, with their health status, appear in Related entities. This includes other services communicating with this service and the infrastructure hosting the service.
- When alerting thresholds are breached, those events appear in the Activity sidebar
By using this information, you can quickly decide whether there's a problem with this service and where you can begin diagnosing the problem.
How OpenTelemetry data shows up (server or client?)
The Summary page shows the golden signals for the server and/or message consumer roles of a service. Other pages offer different views of the service's roles in your distributed system:
- External services shows the service's behavior as a client calling other services, as well as a breakdown of how other services call its endpoints
- Databases shows the service's behavior as a client of databases, specifically
Services can be both servers (responding to requests) and clients (making requests) in the OpenTelemetry data model for tracing. Similarly, services using messaging systems like AWS SQS can be producers and/or consumers of messages. The span.kind attribute specifies the role of the service in a given tracing span.
Metrics or spans
You can choose to use either metrics or spans to power the charts for the golden signals.
Important
When choosing metrics, error rate is displayed only for HTTP servers. The error rate metrics for gRPC or other protocols are not yet shown.
Required attributes
For your OpenTelemetry data to appear in the Summary page, make sure it has the following attributes, in accordance with the OTel semantic conventions:
UI area | Attribute |
|---|---|
Response time |
|
Throughput |
|
Error rate |
|
Instances pane |
|
Distributed tracing insights
You can use the Distributed tracing insights chart to learn whether entities downstream or upstream of your service might be causing increases in the response time, error rate, or throughput of your service.ยท
For example, let's say you want to dig into a spike in the response time of your service when it's related to the external call response time. If distributed tracing recorded a downstream entity causing significantly increased latency for your service around that time, it will show that entity and its change in performance in the list. Click the View trace button to see an example distributed trace where this performance behavior was recorded.
To learn more about distributed tracing data, see Related trace entity signals.
To learn about OpenTelemetry in other UI pages, see Overview of OpenTelemetry data in the UI.