Cisco Cloud Observability ingests metric, event, log, and trace data (MELT) and reports in relation to a Service Instance. If an instance is not uniquely identifiable, the instance reports as an Unknown ID Service Instance and can represent the reporting of multiple logical nodes. This can occur when neither a service instance ID nor container ID is available. Because reporting infrastructure and tags leads to a One:Many relationship and the possible incorrect attribution of load metrics to infrastructure, tags, and version slices, all tags, and infrastructure correlation report as Unknown ID Service Instances.

Cisco Cloud Observability backend is unable to identify a Service Instance when insufficient data is present in the resource attributes of an OpenTelemetry™ packet when a Service Instance ID is not explicitly specified in the resource attribute. 

The semantic conventions for Services are defined in OpenTelemetry Service Specification.

Resource Attribute

Required

service.name

Yes - or default to unknown_service:[optional: process name]

service.version

No

service.instance.id

No

service.namespace

No

Cisco Cloud Observability provides limited fields to map as sufficient information to a service.instance.id.

ServiceInstanceId Supported Values

When will this value be used

service.instance.id

Always if available

container.id

If service.instance.id not available

Unknown ID

If neither service.instance.id or container.id is available


OpenTelemetry™ and Kubernetes® (as applicable) are trademarks of The Linux Foundation®.