Explanation:
Box 1: AzureActivity
The AzureActivity table includes data from many services, including Microsoft Sentinel. To filter in only data from Microsoft Sentinel, start your query with the following code:
Box 2: autocluster()
Example: description: |
'Listing of storage keys is an interesting operation in Azure which might expose additional secrets and PII to callers as well as granting access to VMs. While there are many benign operations of this type, it would be interesting to see if the account performing this activity or the source IP address from which it is being done is anomalous.
The query below generates known clusters of ip address per caller, notice that users which only had single operations do not appear in this list as we cannot learn from it their normal activity (only based on a single event). The activities for listing storage account keys is correlated with this learned clusters of expected activities and activity which is not expected is returned.' AzureActivity
| where OperationNameValue =~ "microsoft.storage/storageaccounts/listkeys/action"
| where ActivityStatusValue == "Succeeded"
| join kind= inner (
AzureActivity
| where OperationNameValue =~ "microsoft.storage/storageaccounts/listkeys/action"
| where ActivityStatusValue == "Succeeded"
| project ExpectedIpAddress=CallerIpAddress, Caller
| evaluate autocluster()
) on Caller
| where CallerIpAddress != ExpectedIpAddress
| summarize StartTime = min(TimeGenerated), EndTime = max(TimeGenerated), ResourceIds = make_set(ResourceId), ResourceIdCount = dcount(ResourceId) by OperationNameValue, Caller, CallerIpAddress
| extend timestamp = StartTime, AccountCustomEntity = Caller, IPCustomEntity = CallerIpAddress Reference:
https://github.com/Azure/Azure-Sentinel/blob/master/Hunting%20Queries/AzureActivity/Anomalous_Listing_O