Advanced query support for AWS and Azure

Understand how to create advanced queries for Service Observability dashboard charts.

Advanced queries in Service Observability support most vendor-specific query languages. Simply copy a query from the observability chart and paste it into the Advanced query field in Service Observability. However, queries to AWS and Azure must be JSON-formatted and follow the guidelines outlined in this topic.

Note:

As of version 1.10, you can import charts directly from AWS and Azure. See Edit data charts on Service Observability dashboard templates for more information.

AWS advanced queries

Data from AWS is accessed using the GetMetricData API. Queries can be either a direct request in JSON format to that API or they can be in the form of a DashboardBody metric widget. For example, you can use the widget representation for a chart that exists on an AWS CloudWatch dashboard.

Direct GetMetricData request

The following example shows a GetMetricData JSON query for a chart that displays the average CPU Utilization for an ELB instance.

{ 

    "StartTime": 1637061900, // can be replaced with “${START}” 
    "EndTime": 1637074500, // can be replaced with “${END}” 
    "MetricDataQueries": [ 
        { 
            "Expression": "SELECT AVG(CPUUtilization) FROM SCHEMA(\"AWS/EC2\", InstanceId)", 
            "Id": "q1", 
            "Period": 300, 
            "Label": "Cluster CpuUtilization" 
        }, 
        { 
            "Id": "m1", 
            "Label": "Unhealthy Behind Load Balancer", 
            "MetricStat": { 
                "Metric": { 
                    "Namespace": "AWS/ApplicationELB", 
                    "MetricName": "UnHealthyHostCount", 
                    "Dimensions": [ 
                        { 
                            "Name": "LoadBalancer", 
                            "Value": "app/EC2Co-EcsEl-EXAMPLE69Q/fdd2210e799e4376" // can be replaced with “${ENTITIES}” or “${ENTITIES_ELB}” 
                        } 
                    ] 
                }, 
                "Period": 300, 
                "Stat": "Average" 
            } 
        } 
    ] 
} 

Copied source for a single object in an AWS CloudWatch dashboard

This example shows the JSON copied directly from a single object in the widgets list of the Source view for a CloudWatch dashboard.

{
  "type": "metric",
  "x": 9,
  "y": 0,
  "width": 15,
  "height": 5,
  "properties": {
    "view": "timeseries",
    "metrics": [
      [ "AWS/ApplicationELB", "RequestCountPerTarget", "TargetGroup", "targetgroup/api-tg/1e3dc9z72fe21ca2", "AvailabilityZone", "us-west-1a" ],
      ["...", "targetgroup/api-tg-cs/82a1db5f950073e1", ".", "." ]
    ],
    "region": "us-west-1"
  }
}

Both types of queries can use template variables to represent a set of entities or the start and end time selected in the UI. For example, instead of querying for a single ELB instance, you can replace the Value field for the LoadBalancer dimension with ${ENTITY_ELB}. The query is run against each ELB instance associated with the selected service in the Service Observability UI. The chart displays a time series for each one.

Note:

JSON objects copied from the source of a CloudWatch dashboard don't include explicit time fields. The values from the time picker in the Service Observability UI are used automatically.

Table 1. Template variables

AWS field	Template variable
Dimensions[n].Value	ENTITIES ENTITIES_HOST ENTITIES_LAMBDA ENTITIES_API_GATEWAY_HTTP ENTITIES_API_GATEWAY_REST ENTITIES_ELB ENTITIES_RDS The Dimensions[n].Value field must be a resource identifier key.
StartTime	START
EndTime	END

Azure advanced queries

You run an advanced query for an Azure-based chart by copying the source from the Azure UI and pasting it in the Advanced query field.

To copy the source:

1. In the Azure UI, export and download the dashboard.
2. Copy the JSON for a single chart. Charts are located at properties.lenses[i].parts[j].metadata.settings.content.options.chart.
3. Paste the JSON into the query field. The JSON must include at least the following:

   ``` 
   { 
   “metrics”: [ 
     “name”: string, 
     “resourceMetadata”: { 
       “id”: string 
       } 
     ] 
   } 
   ``` 

   The id field must be either an Azure resource ID or a template variable.