Azure OpenAI

Collects Azure OpenAI Logs and Metrics


Version	1.4.0 (View all)
Compatible Kibana version(s)	8.13.0 or higher
Supported Serverless project types What's this?	Security Observability
Subscription level What's this?	Basic
Level of support What's this?	Elastic

The Azure OpenAI service provides flexibility to build your own copilot and AI applications. The Azure OpenAI integration collects metrics and logs through azure-monitor and Azure event hub respectively.

To fully populate the Azure OpenAI dashboard lenses, enabling both logs and metrics data streams and setting up the Azure Billing integration in advance is necessary.

Data streams

Logs

The Azure OpenAI logs data stream captures the audit events and the request-response events.

Supported Azure log categories:

Data Stream	Log Category
logs	Audit
logs	RequestResponse
logs	ApiManagementGatewayLogs

Requirements and setup

Refer to the Azure Logs page for more information about setting up and using this integration.

Default Logging

The Azure OpenAI provides native logging and monitoring with which you can track the telemetry of the service. The Audit and the RequestResponse log categories comes under the native logging. But the default logging doesn't log the inputs and outputs of the service. These are the useful to ensure that the services operates as expected.

The logs collected using the default cognitive services is listed here.

Advance Logging

The API Management services provides the advance logging capabilities. The ApiManagementGatewayLogs category comes under the advance logging. This is not directly available in the Azure OpenAI service itself. You have to setup the API Management services in the Azure to access the Azure OpenAI. Once the setup is done add the diagnostic setting for the API Management service.

You can find information on how to implement the comprehensive solution using API Management services to monitor the Azure OpenAI services here.

Diagnostic settings

Enable the category Logs related to ApiManagement Gateway to stream the logs to the event hub.

   ┌──────────────────┐      ┌──────────────┐     ┌─────────────────┐
   │   APIM service   │      │  Diagnostic  │     │    Event Hub    │
   │    <<source>>    │─────▶│   settings   │────▶│ <<destination>> │
   └──────────────────┘      └──────────────┘     └─────────────────┘

The logs collected using the API Management services for the enterprise customer of the Azure OpenAI services is listed here. This records the inputs and outputs of the request, like prompts, tokens, and model usage.

Settings

Refer to the settings section for more details on the configuration.

Logs Reference

An example event for logs looks as following:

{
    "@timestamp": "2024-04-08T12:23:02.435Z",
    "azure": {
        "open_ai": {
            "caller_ip_address": "81.2.69.***",
            "category": "RequestResponse",
            "correlation_id": "9d3a6e98-fc11-48d0-82cf-4de065c1a1f8",
            "event": "ShoeboxCallResult",
            "location": "eastus",
            "operation_name": "Create_Thread",
            "properties": {
                "api_name": "Azure OpenAI API version 2024-02-15-preview",
                "object_id": "",
                "request_length": 2,
                "request_time": 638481757794854611,
                "response_length": 113,
                "response_time": 638481757795877942
            },
            "result_signature": "200",
            "tenant": "eastus"
        },
        "resource": {
            "group": "obs-openai-service-rs",
            "id": "/subscriptions/12cabcb4-86e8-404f-a3d2-1dc9982f45ca/resourcegroups/obs-openai-service-rs/providers/microsoft.cognitiveservices/accounts/obs-openai-test-01",
            "name": "obs-openai-test-01",
            "provider": "microsoft.cognitiveservices/accounts"
        }
    },
    "cloud": {
        "provider": "azure"
    },
    "event": {
        "duration": 102000000,
        "original": "{\"Tenant\":\"eastus\",\"callerIpAddress\":\"81.2.69.***\",\"category\":\"RequestResponse\",\"correlationId\":\"9d3a6e98-fc11-48d0-82cf-4de065c1a1f8\",\"durationMs\":102,\"event\":\"ShoeboxCallResult\",\"location\":\"eastus\",\"operationName\":\"Create_Thread\",\"properties\":\"{\\\"apiName\\\":\\\"Azure OpenAI API version 2024-02-15-preview\\\",\\\"requestTime\\\":638481757794854611,\\\"requestLength\\\":2,\\\"responseTime\\\":638481757795877942,\\\"responseLength\\\":113,\\\"objectId\\\":\\\"\\\"}\",\"resourceId\":\"/SUBSCRIPTIONS/12CABCB4-86E8-404F-A3D2-1DC9982F45CA/RESOURCEGROUPS/OBS-OPENAI-SERVICE-RS/PROVIDERS/MICROSOFT.COGNITIVESERVICES/ACCOUNTS/OBS-OPENAI-TEST-01\",\"resultSignature\":\"200\",\"time\":\"2024-04-08T12:23:02.4350000Z\"}"
    },
    "tags": [
        "preserve_original_event"
    ]
}

ECS Field Reference

Please refer to the following document for detailed information on ECS fields.

Exported fields

Field	Description	Type
@timestamp	Event timestamp.	date
azure.open_ai.asset_identity	The asset identity key.	keyword
azure.open_ai.caller_ip_address	The client IP address. (x - last octet masked).	keyword
azure.open_ai.category	The log category name.	keyword
azure.open_ai.correlation_id	The correlation id as key.	keyword
azure.open_ai.deployment_version	The deployment version.	keyword
azure.open_ai.event	The event type of the service request.	keyword
azure.open_ai.is_request_success	True if the request is success else return false.	boolean
azure.open_ai.location	The location.	keyword
azure.open_ai.operation_name	The log action performed.	keyword
azure.open_ai.properties.api_id	The request API Id.	keyword
azure.open_ai.properties.api_name	The API name of the request.	keyword
azure.open_ai.properties.api_revision	The request API revision.	keyword
azure.open_ai.properties.backend_method	The backend request method.	keyword
azure.open_ai.properties.backend_protocol	The backend protocol.	keyword
azure.open_ai.properties.backend_request_body.messages.content	The prompt input.	keyword
azure.open_ai.properties.backend_request_body.messages.role	The API access role.	keyword
azure.open_ai.properties.backend_request_body.model	The model name.	keyword
azure.open_ai.properties.backend_response_body.choices.content_filter_results	Content filtered by custom content filter.	flattened
azure.open_ai.properties.backend_response_body.choices.finish_reason	A string indicating the reason why the response was generated (e.g., "max_tokens").	keyword
azure.open_ai.properties.backend_response_body.choices.index	The index of the response in the array.	long
azure.open_ai.properties.backend_response_body.choices.logprobs	An object containing information about the probability distribution over possible responses.	flattened
azure.open_ai.properties.backend_response_body.choices.message.content	The response text content.	keyword
azure.open_ai.properties.backend_response_body.choices.message.role	The API access role.	keyword
azure.open_ai.properties.backend_response_body.created	The timestamp when the request was created.	long
azure.open_ai.properties.backend_response_body.error.code	The error code.	keyword
azure.open_ai.properties.backend_response_body.error.innererror.code	The error code.	keyword
azure.open_ai.properties.backend_response_body.error.innererror.content_filter_result	Content filtered by custom content filter.	flattened
azure.open_ai.properties.backend_response_body.error.message	The error message.	text
azure.open_ai.properties.backend_response_body.error.param	Parameter passed to the API.	keyword
azure.open_ai.properties.backend_response_body.error.status	The response status code.	long
azure.open_ai.properties.backend_response_body.id	A unique identifier for the request.	keyword
azure.open_ai.properties.backend_response_body.model	The ID of the OpenAI model used to generate the response.	keyword
azure.open_ai.properties.backend_response_body.object	The operation type.	keyword
azure.open_ai.properties.backend_response_body.prompt_filter_results.content_filter_results	Content filtered by custom content filter.	flattened
azure.open_ai.properties.backend_response_body.prompt_filter_results.prompt_index	Index of the prompt used to generate response.	long
azure.open_ai.properties.backend_response_body.system_fingerprint	The fingerprint is generated by Azure API Management and is used to track the performance and usage of the backend service.	keyword
azure.open_ai.properties.backend_response_body.usage.input_tokens	the total input tokens.	long
azure.open_ai.properties.backend_response_body.usage.output_tokens	The total output tokens.	long
azure.open_ai.properties.backend_response_body.usage.total_tokens	The sum of input and output tokens.	long
azure.open_ai.properties.backend_response_code	The backend HTTP response code.	long
azure.open_ai.properties.backend_time	The backend response time.	long
azure.open_ai.properties.backend_url	The backend URL connects to the Azure OpenAI model.	keyword
azure.open_ai.properties.cache	The request cache.	keyword
azure.open_ai.properties.client_protocol	The client HTTP protocol.	keyword
azure.open_ai.properties.client_tls_version	The client TLS version.	keyword
azure.open_ai.properties.model_deployment_name	The deployed model name.	keyword
azure.open_ai.properties.model_name	The OpenAI model.	keyword
azure.open_ai.properties.model_version	The OpenAI model version.	keyword
azure.open_ai.properties.object_id	The object id of the request.	keyword
azure.open_ai.properties.operation_id	The operation performed.	keyword
azure.open_ai.properties.request_length	Length of the request.	double
azure.open_ai.properties.request_time	Request time taken.	long
azure.open_ai.properties.response_length	Length of the response.	double
azure.open_ai.properties.response_time	Response time taken.	long
azure.open_ai.properties.stream_type	The stream type of the request.	keyword
azure.open_ai.result_signature	The response status.	keyword
azure.open_ai.sku	Stock Keeping Unit that is associated with a particular API Management instance.	keyword
azure.open_ai.tenant	The tenant location.	keyword
azure.open_ai.truncated	Condition where the response message is too large to fit in a single packet, so it is truncated or cut off.	long
azure.resource.authorization_rule	Authorization rule	keyword
azure.resource.group	The resource group	keyword
azure.resource.id	Resource ID	keyword
azure.resource.name	The name of the resource	keyword
azure.resource.namespace	Resource type/namespace	keyword
azure.resource.provider	The resource group	keyword
azure.resource.type	The type of the resource	keyword
azure.subscription_id	The subscription ID	keyword
data_stream.dataset	Data stream dataset.	constant_keyword
data_stream.namespace	Data stream namespace.	constant_keyword
data_stream.type	Data stream type.	constant_keyword

Metrics

The metrics data stream collects the cognitive service metrics that is specific to the Azure OpenAI service. The metrics that are specific to PTUs will only be available with the provisioned deployments.

Refer here to find more details on the metrics applicable to PTU only deployment.

Requirements

Before you start, check the Authentication and costs section.

Setup

Follow these step-by-step instructions on how to set up an Azure metrics integration.

Data stream specific configuration notes

Period:: (string) Reporting interval. Metrics will have a timegrain of 5 minutes, so the Period configuration option for azure_openai should have a value of 300s or multiple of 300sfor relevant results.

Resource IDs:: ([]string) The fully qualified ID's of the resource, including the resource name and resource type. Has the format /subscriptions/{guid}/resourceGroups/{resource-group-name}/providers/{resource-provider-namespace}/{resource-type}/{resource-name}. Should return a list of resources.

Resource Groups:: ([]string) This option will return all Azure OpenAI services inside the resource group.

If no resource filter is specified, then all Azure OpenAI services inside the entire subscription will be considered.

The primary aggregation value will be retrieved for all the metrics contained in the namespaces. The aggregation options are avg, sum, min, max, total, count.

Metrics Reference

An example event for metrics looks as following:

{
    "@timestamp": "2024-04-11T01:46:00.000Z",
    "agent": {
        "ephemeral_id": "dcff0e53-fadb-4e97-86a2-1e611f12fc34",
        "id": "0c09f097-dc30-44c5-b3e7-083f1a14873c",
        "name": "docker-fleet-agent",
        "type": "metricbeat",
        "version": "8.13.0"
    },
    "azure": {
        "dimensions": {
            "api_name": "Azure OpenAI API version 2024-04-01-preview",
            "model_deployment_name": "gpt-chat-pilot",
            "model_name": "gpt-35-turbo",
            "model_version": "0301",
            "operation_name": "ChatCompletions_Create",
            "region": "East US",
            "status_code": "200",
            "stream_type": "Streaming"
        },
        "namespace": "Microsoft.CognitiveServices/accounts",
        "open_ai": {
            "requests": {
                "total": 1
            }
        },
        "resource": {
            "group": "obs-openai-service-rs",
            "id": "/subscriptions/12cabcb4-86e8-404f-a3d2-1dc9982f45ca/resourceGroups/obs-openai-service-rs/providers/Microsoft.CognitiveServices/accounts/obs-openai-test-01",
            "name": "obs-openai-test-01",
            "type": "Microsoft.CognitiveServices/accounts"
        },
        "subscription_id": "12cabcb4-86e8-404f-a3d2-1dc9982f45ca",
        "timegrain": "PT1M"
    },
    "cloud": {
        "provider": "azure",
        "region": "eastus"
    },
    "data_stream": {
        "dataset": "azure.open_ai",
        "namespace": "default",
        "type": "metrics"
    },
    "ecs": {
        "version": "8.11.0"
    },
    "elastic_agent": {
        "id": "0c09f097-dc30-44c5-b3e7-083f1a14873c",
        "snapshot": false,
        "version": "8.13.0"
    },
    "event": {
        "agent_id_status": "verified",
        "dataset": "azure.open_ai",
        "duration": 2216811793,
        "ingested": "2024-04-11T01:52:30Z",
        "module": "azure"
    },
    "metricset": {
        "name": "monitor",
        "period": 300000
    },
    "service": {
        "type": "azure"
    }
}