Abnormal Deny Rate for Source IP

Status: available
Severity: medium
Time window: 25h
Group by: Fqdn, SourceIp
Source: github.com/Azure/Azure-Sentinel

'Identifies abnormal deny rate for specific source IP to destination IP based on the normal average and standard deviation learned during a configured period. This can indicate potential exfiltration, initial access or C2, where attacker tries to exploit the same vulnerability on machines in the organization, but is being blocked by firewall rules. Configurable Parameters: - Minimum of stds threshold - the number of stds to use in the threshold calculation. Default is set to 3. - Learning period time - learning period for threshold calculation in days. Default is set to 5. - Bin time - learning buckets time in hours. Default is set to 1 hour. - Minimum threshold - minimum threshold for alert. Default is set to 5. - Minimum bucket threshold - minimum learning buckets threshold for alert. Default is set to 5.'

MITRE ATT&CK coverage

Tactic	Techniques
Initial Access	`T1190` Exploit Public-Facing Application
Discovery	`T1046` Network Service Discovery
Command & Control	`T1568.001` Dynamic Resolution: Fast Flux DNS, `T1568.002` Dynamic Resolution: Domain Generation Algorithms
Exfiltration	`T1041` Exfiltration Over C2 Channel

Rule body kusto

id: d36bb1e3-5abc-4037-ad9a-24ba3469819e
name: Abnormal Deny Rate for Source IP
description: |
  'Identifies abnormal deny rate for specific source IP to destination IP based on the normal average and standard deviation learned during a configured period. This can indicate potential exfiltration, initial access or C2, where attacker tries to exploit the same vulnerability on machines in the organization, but is being blocked by firewall rules.
  
  Configurable Parameters:
  
  - Minimum of stds threshold - the number of stds to use in the threshold calculation. Default is set to 3.
  - Learning period time - learning period for threshold calculation in days. Default is set to 5.
  - Bin time - learning buckets time in hours. Default is set to 1 hour.
  - Minimum threshold - minimum threshold for alert. Default is set to 5.
  - Minimum bucket threshold - minimum learning buckets threshold for alert. Default is set to 5.'
severity: Medium
status: Available
requiredDataConnectors:
  - connectorId: AzureFirewall
    dataTypes: 
      - AzureDiagnostics
      - AZFWApplicationRule
      - AZFWNetworkRule
      - AZFWFlowTrace
      - AZFWIdpsSignature
queryFrequency: 1h
queryPeriod: 25h
triggerOperator: gt
triggerThreshold: 0
tactics:
  - InitialAccess
  - Exfiltration
  - CommandAndControl
  - Discovery
relevantTechniques:
  - T1190
  - T1041
  - T1568.001
  - T1568.002
  - T1046
query: |
  let LearningPeriod = 1d;
  let RunTime = 1h;
  let StartLearningPeriod = LearningPeriod + RunTime;
  let EndRunTime = RunTime - 1h;
  let BinTime = 1h;
  let NumOfStdsThreshold = 3;
  let MinThreshold = 5.0;
  let MinLearningBuckets = 5;
  // Restrict unioned tables to the full analysis window to avoid full-table scans
  let FullWindowStart = ago(StartLearningPeriod);
  let FullWindowEnd = now();
  let TrafficLogs = (union isfuzzy=true
  (AzureDiagnostics
  | where TimeGenerated between (FullWindowStart .. FullWindowEnd)
  | where OperationName == "AzureFirewallApplicationRuleLog" or OperationName == "AzureFirewallNetworkRuleLog" 
  | parse msg_s with * "from " SourceIp ":" SourcePort:int " to " Fqdn ":" DestinationPort:int ". " * "Action: " Action "." *
  | where Action == "Deny"
  | where isnotempty(Fqdn) and isnotempty(SourceIp)),
  (AZFWNetworkRule
  | where TimeGenerated between (FullWindowStart .. FullWindowEnd)
  | extend Fqdn = DestinationIp
  | where Action == "Deny"
  | where isnotempty(Fqdn) and isnotempty(SourceIp)),
  (AZFWFlowTrace
  | where TimeGenerated between (FullWindowStart .. FullWindowEnd)
  | extend Fqdn = DestinationIp
  | where Action == "Deny"
  | where isnotempty(Fqdn) and isnotempty(SourceIp)),
  (AZFWIdpsSignature
  | where TimeGenerated between (FullWindowStart .. FullWindowEnd)
  | extend Fqdn = DestinationIp
  | where Action == "Deny"
  | where isnotempty(Fqdn) and isnotempty(SourceIp)),
  (AZFWApplicationRule
  | where TimeGenerated between (FullWindowStart .. FullWindowEnd)
  | where Action == "Deny"
  | where isnotempty(Fqdn) and isnotempty(SourceIp)));
  let LearningSrcIpDenyRate = (TrafficLogs
  | where TimeGenerated between (ago(StartLearningPeriod) .. ago(RunTime))
  | summarize count() by SourceIp, bin(TimeGenerated, BinTime), Fqdn
  | summarize LearningTimeSrcIpDenyRateAvg = avg(count_), LearningTimeSrcIpDenyRateStd = stdev(count_), LearningTimeBuckets = count() by SourceIp, Fqdn
  | where LearningTimeBuckets > MinLearningBuckets);
  let AlertTimeSrcIpDenyRate = (TrafficLogs
  | where TimeGenerated between (ago(RunTime) .. ago(EndRunTime))
  | summarize AlertTimeSrcIpDenyRateCount = count() by SourceIp, Fqdn);
  AlertTimeSrcIpDenyRate
  | join kind=leftouter (LearningSrcIpDenyRate) on SourceIp, Fqdn
  | extend LearningThreshold = max_of(LearningTimeSrcIpDenyRateAvg + NumOfStdsThreshold * LearningTimeSrcIpDenyRateStd, MinThreshold)
  | where AlertTimeSrcIpDenyRateCount > LearningThreshold
  | project SourceIp, Fqdn, AlertTimeSrcIpDenyRateCount, LearningTimeBuckets, LearningThreshold
entityMappings:
  - entityType: IP
    fieldMappings:
      - identifier: Address
        columnName: SourceIp
  - entityType: URL
    fieldMappings:
      - identifier: Url
        columnName: Fqdn
customDetails:
  AlertCount: AlertTimeSrcIpDenyRateCount
  Threshold: LearningThreshold
alertDetailsOverride:
  alertDisplayNameFormat: 'Abnormal Deny Rate for Source IP {{SourceIp}}'
  alertDescriptionFormat: |
    The source IP {{SourceIp}} has an abnormal deny rate to destination {{Fqdn}} with {{AlertTimeSrcIpDenyRateCount}} denies, which is above the calculated threshold. This could indicate potential exfiltration, initial access or C2 activities.
version: 1.1.3
kind: Scheduled

Stages and Predicates

Parameters

let LearningPeriod = 1d;
let RunTime = 1h;
let StartLearningPeriod = LearningPeriod + RunTime;
let EndRunTime = RunTime - 1h;
let BinTime = 1h;
let NumOfStdsThreshold = 3;
let MinThreshold = 5.0;
let MinLearningBuckets = 5;
let FullWindowStart = ago(StartLearningPeriod);
let FullWindowEnd = now();

Let binding: `LearningSrcIpDenyRate`

let LearningSrcIpDenyRate = (TrafficLogs
| where TimeGenerated between (ago(StartLearningPeriod) .. ago(RunTime))
| summarize count() by SourceIp, bin(TimeGenerated, BinTime), Fqdn
| summarize LearningTimeSrcIpDenyRateAvg = avg(count_), LearningTimeSrcIpDenyRateStd = stdev(count_), LearningTimeBuckets = count() by SourceIp, Fqdn
| where LearningTimeBuckets > MinLearningBuckets);

Derived from RunTime, StartLearningPeriod, BinTime, MinLearningBuckets, TrafficLogs.

The stages below define let AlertTimeSrcIpDenyRate (the rule's main pipeline source).

Stage 1: `union`

union isfuzzy=true

Stage 2: `source`

AzureDiagnostics

Stage 3: `where`

| where TimeGenerated between (FullWindowStart .. FullWindowEnd)

The stages below run on AlertTimeSrcIpDenyRate (the outer pipeline).

Stage 4: `where`

AlertTimeSrcIpDenyRate
| where OperationName == "AzureFirewallApplicationRuleLog" or OperationName == "AzureFirewallNetworkRuleLog"

Stage 5: `parse`

| parse msg_s with * "from " SourceIp ":" SourcePort:int " to " Fqdn ":" DestinationPort:int ". " * "Action: " Action "." *

Stage 6: `where`

| where Action == "Deny"

Stage 7: `where`

| where isnotempty(Fqdn) and isnotempty(SourceIp)

Stage 8: `source`

AZFWNetworkRule

Stage 9: `where`

| where TimeGenerated between (FullWindowStart .. FullWindowEnd)

Stage 10: `extend`

| extend Fqdn = DestinationIp

Stage 11: `where`

| where Action == "Deny"

Stage 12: `where`

| where isnotempty(Fqdn) and isnotempty(SourceIp)

Stage 13: `source`

AZFWFlowTrace

Stage 14: `where`

| where TimeGenerated between (FullWindowStart .. FullWindowEnd)

Stage 15: `extend`

| extend Fqdn = DestinationIp

Stage 16: `where`

| where Action == "Deny"

Stage 17: `where`

| where isnotempty(Fqdn) and isnotempty(SourceIp)

Stage 18: `source`

AZFWIdpsSignature

Stage 19: `where`

| where TimeGenerated between (FullWindowStart .. FullWindowEnd)

Stage 20: `extend`

| extend Fqdn = DestinationIp

Stage 21: `where`

| where Action == "Deny"

Stage 22: `where`

| where isnotempty(Fqdn) and isnotempty(SourceIp)

Stage 23: `source`

AZFWApplicationRule

Stage 24: `where`

| where TimeGenerated between (FullWindowStart .. FullWindowEnd)

Stage 25: `where`

| where Action == "Deny"

Stage 26: `where`

| where isnotempty(Fqdn) and isnotempty(SourceIp)

Stage 27: `where`

| where TimeGenerated between (ago(RunTime) .. ago(EndRunTime))

Stage 28: `summarize`

| summarize AlertTimeSrcIpDenyRateCount = count() by SourceIp, Fqdn

Threshold: gt LearningThreshold

Stage 29: `join`

| join kind=leftouter (LearningSrcIpDenyRate) on SourceIp, Fqdn

Stage 30: `extend`

| extend LearningThreshold = max_of(LearningTimeSrcIpDenyRateAvg + NumOfStdsThreshold * LearningTimeSrcIpDenyRateStd, MinThreshold)

Stage 31: `where`

| where AlertTimeSrcIpDenyRateCount > LearningThreshold

Stage 32: `project`

| project SourceIp, Fqdn, AlertTimeSrcIpDenyRateCount, LearningTimeBuckets, LearningThreshold

Indicators

Each row is a field, operator, and value that the rule matches. The corpus column counts how many other rules in the catalog look for the same combination: high numbers point to widely-used, community-vetted indicators. Blank or 1 shows that the indicator is specific to this rule.

Field	Kind	Values
`Action`	eq	`Deny` transforms: `cased`
`AlertTimeSrcIpDenyRateCount`	gt	`LearningThreshold` transforms: `cased`
`Fqdn`	is_not_null	(no value, null check)
`LearningTimeBuckets`	gt	`5` transforms: `cased`
`OperationName`	eq	`AzureFirewallApplicationRuleLog` transforms: `cased` `AzureFirewallNetworkRuleLog` transforms: `cased`
`SourceIp`	is_not_null	(no value, null check)
`TimeGenerated`	ge	`FullWindowStart`
`TimeGenerated`	le	`FullWindowEnd`

Output fields

Fields the rule emits when it matches. Chronicle authors list these in the outcome block; they appear on the detection and $risk_score drives alerting. Sentinel / Defender XDR rules build them up through project / summarize / extend stages. Sentinel maps these into alert fields via entityMappings and customDetails; Defender XDR custom detections surface them as alert fields directly.

Field	Source
`AlertTimeSrcIpDenyRateCount`	`project`
`Fqdn`	`project`
`LearningThreshold`	`project`
`LearningTimeBuckets`	`project`
`SourceIp`	`project`

`j` / `k`	Scroll down / up
`d` / `u`	Half-page down / up
`gg` / `G`	Top / bottom
`h` / `l`	History back / forward
`f`	Follow link (`Shift` = new tab)
`/`	Focus search
`?`	Toggle this help
`↑` / `↓`	Navigate search results
`Enter`	Open highlighted result
`Esc`	Close results / dialog

`type:`	`events` / `rules` / `providers`
`vendor:`	`sigma` / `elastic` / `splunk` / `kusto` / `chronicle` (vendor name alone also works: `sigma:`, `kql:`, `secops:`…)
`tactic:`	TA-id, slug, or name: `credential_access`, `TA0006`
`technique:`	technique or sub-technique ID: `T1003`, `T1003.001` (alias `tech:`)
`severity:`	`critical` / `high` / `medium` / `low` / `informational` (alias `sev:`)
`risk_score`	Numeric comparison on the Elastic risk score (0 to 100): `risk_score>50`, `risk_score<=20`, `risk_score=99` (alias `risk`; Elastic rules only)
`stages:`	Rules with exactly N pipeline stages
`correlation:`	`single_event` / `sequence` / `alternatives` / `alternatives_cross_log` / `all_required` / `correlated`
`with:`	Co-occurrence event-id; stacks (`with:4624 with:4769`) to require all, while a comma list in one occurrence (`with:4624,4769`) is an either-or group. Implies multi-event
`like:`	Structural neighbors of a rule slug (equivalents + subsumption stricter / broader): `like:comsvcs_lsass_memory_dump-splunk-sysmon`
`groupby:`	Entity-grouping substring match against `group_by_keys`: `groupby:user`, `groupby:host`
`uses:`	Rules whose predicate tree touches the field (any kind, any value): `uses:CommandLine`
`excludes:`	Rules with top-level `not()` clauses on the field (FP whitelists): `excludes:ParentImage`
`field:` / `value:`	Predicate search; narrows rule cards to those with a matching leaf and drives the indicator tier. Unquoted = substring, wildcards allowed (`value:mimikatz`)
`indicator:`	Shorthand for `field:F value:V`: `indicator:Image=*\powershell.exe`
`kind:`	Filter by predicate kind. Narrows rule cards to those carrying a matching predicate leaf (`vendor:elastic kind:cidr_match`) and drives the indicator tier: `contains` / `starts_with` / `ends_with` / `regex` / `cidr` / `eq` / `in` … (operator aliases `op:`/`match:`)
`has:` / `no:`	`sample`, `field`, `notes`, `refs`, `trace`, `thirdparty`, `rule`, `pattern`, `timewindow`, `threshold`, `newterms`, `sigma`/`elastic`/`splunk`/`kusto`/`chronicle`
`-op:val`	Exclude matches; works on most operators but not `type:`/`like:`/`has:`/`no:` (use `no:<flag>` to exclude a rule flag): `tactic:execution -vendor:splunk`. Standalone `-kind:`/`-field:`/`-value:` drop every rule carrying a matching predicate leaf (`type:rules -kind:is_null`)
`field:"…"` / `value:"…"`	Quoted value = anchored exact match (also allows spaces): `value:"net user"`
`a,b`	Comma = OR inside one operator (`vendor:sigma,elastic`, `severity:high,critical`); repeating a facet merges the same way. `field:`/`value:` never split (literal commas)
`vendors:` / `stage:`	Singular and plural spellings fold to the canonical operator and value: `tactics:` = `tactic:`, `type:event` = `type:events`, `correlation:sequences` = `correlation:sequence`, `has:thresholds` = `has:threshold`
`"quoted phrase"`	Exact-match a multi-word phrase (free text)

Abnormal Deny Rate for Source IP

MITRE ATT&CK coverage

Rule body kusto

Stages and Predicates

Parameters

Let binding: LearningSrcIpDenyRate

Stage 1: union

Stage 2: source

Stage 3: where

Stage 4: where

Stage 5: parse

Stage 6: where

Stage 7: where

Stage 8: source

Stage 9: where

Stage 10: extend

Stage 11: where

Stage 12: where

Stage 13: source

Stage 14: where

Stage 15: extend

Stage 16: where

Stage 17: where

Stage 18: source

Stage 19: where

Stage 20: extend

Stage 21: where

Stage 22: where

Stage 23: source

Stage 24: where

Stage 25: where

Stage 26: where

Stage 27: where

Stage 28: summarize

Stage 29: join

Stage 30: extend

Stage 31: where

Stage 32: project