편집

다음을 통해 공유

Access Microsoft Graph activity logs

  • 아티클

Microsoft Graph activity logs are an audit trail of all HTTP requests that the Microsoft Graph service received and processed for a tenant. Tenant administrators can enable the collection and configure downstream destinations for these logs using diagnostic settings in Azure Monitor. The logs are stored in Log Analytics for analysis; you can export them to Azure Storage for long-term storage or stream with Azure Event Hubs to external SIEM tools for alerting, analysis, or archival.

All logs for API requests made from line of business applications, API clients, SDKs, and by Microsoft applications like Outlook, Microsoft Teams, or the Microsoft Entra admin center are available.

This service is available in the following national cloud deployments.

Global service US Government L4 US Government L5 (DOD) China operated by 21Vianet
✅ ✅ ✅ ❌

Prerequisites

To access the Microsoft Graph activity logs, you need the following privileges.

  • A Microsoft Entra ID P1 or P2 tenant license in your tenant.
  • An administrator with a supported Microsoft Entra administrator role. Security Administrator is the only least privileged admin role supported for configuring diagnostic settings.
  • An Azure subscription with one of the following log destinations are configured, and permissions to access data in the corresponding log destinations.
    • An Azure Log Analytics workspace to send logs to Azure Monitor
    • An Azure Storage Account for which you have List Keys permissions
    • An Azure Event Hubs namespace to integrate with third-party solutions

What data is available in the Microsoft Graph activity logs?

The following data relating to API requests is available for Microsoft Graph activity logs on the Logs Analytics interface.

Column Type Description
AadTenantId string The Azure AD tenant ID.
ApiVersion string The API version of the event.
AppId string The identifier for the application.
ATContent string Reserved for future use.
ATContentH string Reserved for future use.
ATContentP string Reserved for future use.
_BilledSize real The record size in bytes
ClientAuthMethod int Indicates how the client was authenticated. For a public client, the value is 0. If client ID and client secret are used, the value is 1. If a client certificate was used for authentication, the value is 2.
ClientRequestId string Optional. The client request identifier when sent. If no client request identifier is sent, the value will be equal to the operation identifier.
DurationMs int The duration of the request in milliseconds.
IdentityProvider string The identity provider that authenticated the subject of the token.
IPAddress string The IP address of the client from where the request occurred.
_IsBillable string Specifies whether ingesting the data is billable. When _IsBillable is false ingestion isn't billed to your Azure account
Location string The name of the region that served the request.
OperationId string The identifier for the batch. For non-batched requests, this will be unique per request. For batched requests, this will be the same for all requests in the batch.
RequestId string The identifier representing the request.
RequestMethod string The HTTP method of the event.
RequestUri string The URI of the request.
ResponseSizeBytes int The size of the response in Bytes.
ResponseStatusCode int The HTTP response status code for the event.
Roles string The roles in token claims.
Scopes string The scopes in token claims.
ServicePrincipalId string The identifier of the servicePrincipal making the request.
SignInActivityId string The identifier representing the sign-in activitys.
SourceSystem string The type of agent the event was collected by. For example, OpsManager for Windows agent, either direct connect or Operations Manager, Linux for all Linux agents, or Azure for Azure Diagnostics
TenantId string The Log Analytics workspace ID
TimeGenerated datetime The date and time the request was received.
TokenIssuedAt datetime The timestamp the token was issued at.
Type string The name of the table
UserAgent string The user agent information related to request.
UserId string The identifier of the user making the request.
Wids string Denotes the tenant-wide roles assigned to this user.

Common use cases for Microsoft Graph activity logs

  • Get full visibility into the transactions made by applications and other API clients that you have consented to in the tenant.
  • Identify the activities that a compromised user account conducted in your tenant.
  • Build detections and behavioral analysis to identify suspicious or anomalous use of Microsoft Graph APIs.
  • Investigate unexpected or suspicious privileged assignment of application permissions.
  • Identify problematic or unexpected behaviors for client applications such as extreme call volumes.
  • Correlate Microsoft Graph requests made by a user or app with sign-in information.

Configure to receive the Microsoft Graph activity logs

You can configure to stream the logs through the Diagnostic Setting in the Azure portal or through Azure Resource Manager APIs. For more information, see the guidance in the following articles:

The following articles guide you to configure the storage destinations:

Cost planning estimates

If you already have a Microsoft Entra ID P1 license, you need an Azure subscription to set up the Log Analytics workspace, Storage account, or Event Hubs. The Azure subscription comes at no cost, but you have to pay to utilize Azure resources.

The amount of data logged and, thus, the cost incurred, can vary significantly depending on the tenant size and the applications in your tenant that interact with Microsoft Graph APIs. The following table provides some estimates for log data size to aid the price calculation. Use these estimations for general consideration only.

Users in tenant Storage GiB/month Event Hubs Messages/month Azure Monitor Logs GiB/month
1000 14 62K 15
100000 1000 4.8M 1200

See the following pricing calculations for respective services:

Cost reduction for Log Analytics

If you're ingesting the logs to a Log Analytics Workspace but are only interested in logs filtered by a criteria, such as omitting certain columns or rows, you can partially reduce costs by applying a workspace transformation on the Microsoft Graph Activity Logs table. To find out more about workspace transformations, how it affects ingestion costs, and how to apply a transformation to your Microsoft Graph Activity Logs, see Data collection transformations in Azure Monitor.

An alternative approach to reduce Log Analytics cost is to switch to the Basic log data plan which lowers the bills by providing reduced capabilities. For more information, see Set a table's log data plan to Basic or Analytics.

Azure Monitor Logs query examples

If you send Microsoft Graph activity logs to a Log Analytics workspace, you can query the logs using Kusto Query Language (KQL). For more information about queries in Log Analytics Workspace, see Analyze Microsoft Entra activity logs with Log Analytics. You can use these queries for data exploration, to build alert rules, build Azure dashboards, or integrate into your custom applications using the Azure Monitor Logs API or Query SDK.

The following Kusto query identifies the top 20 entities making requests to groups resources that are failing due to authorization:

MicrosoftGraphActivityLogs
| where TimeGenerated >= ago(3d)
| where ResponseStatusCode == 401 or ResponseStatusCode == 403 
| where RequestUri contains "/groups"
| summarize UniqueRequests=count_distinct(RequestId) by AppId, ServicePrincipalId, UserId
| sort by UniqueRequests desc
| limit 20

The following Kusto query identifies resources queried or modified by potentially risky users:

MicrosoftGraphActivityLogs
| where TimeGenerated > ago(30d)
| join AADRiskyUsers on $left.UserId == $right.Id
| extend resourcePath = replace_string(replace_string(replace_regex(tostring(parse_url(RequestUri).Path), @'(\/)+','/'),'v1.0/',''),'beta/','')
| summarize RequestCount=dcount(RequestId) by UserId, RiskState, resourcePath, RequestMethod, ResponseStatusCode

The following Kusto query allows you to correlate the Microsoft Graph activity logs and sign-in logs. Activity logs from Microsoft applications may not all have matching sign-in log entries. For more information, see Sign-in logs known limitations.

MicrosoftGraphActivityLogs
| where TimeGenerated > ago(7d)
| join kind=leftouter (union SigninLogs, AADNonInteractiveUserSignInLogs, AADServicePrincipalSignInLogs, AADManagedIdentitySignInLogs, ADFSSignInLogs
    | where TimeGenerated > ago(7d))
    on $left.SignInActivityId == $right.UniqueTokenIdentifier

The following Kusto query identifies apps that are getting throttled:

MicrosoftGraphActivityLogs 
| where TimeGenerated > ago(3d) 
| where ResponseStatusCode == 429 
| extend path = replace_string(replace_string(replace_regex(tostring(parse_url(RequestUri).Path), @'(\/)+','//'),'v1.0/',''),'beta/','') 
| extend UriSegments =  extract_all(@'\/([A-z2]+|\$batch)($|\/|\(|\$)',dynamic([1]),tolower(path)) 
| extend OperationResource = strcat_array(UriSegments,'/')| summarize RateLimitedCount=count() by AppId, OperationResource, RequestMethod 
| sort by RateLimitedCount desc 
| limit 100

The following query allows you to render a time-series chart:

MicrosoftGraphActivityLogs 
| where TimeGenerated  between (ago(3d) .. ago(1h))  
| summarize EventCount = count() by bin(TimeGenerated, 10m) 
| render timechart 
    with ( 
    title="Recent traffic patterns", 
    xtitle="Time", 
    ytitle="Requests", 
    legend=hidden 
    )

Limitations

  • The Microsoft Graph activity logs feature allows the tenant administrators to collect logs for the resource tenant. This feature doesn't allow you to see the activities of a multitenant application in another tenant.
  • You can't filter Microsoft Graph activity logs through diagnostic settings in Azure Monitor. However, options are available to reduce costs in Azure Log Analytics Workspace. For more information, see Workspace transformation.
  • In most regions, the events are available and delivered to the configuration destination within 30 minutes. In less common cases, some events might take up to 2 hours to be delivered to the destination.

Related content