Detection rules › Kusto

Potential DGA(Domain Generation Algorithm) detected via Repetitive Failures - Anomaly based (ASIM DNS Solution)

Status
available
Severity
medium
Time window
14d
Group by
SrcIpAddr, exists, sumexist
Source
github.com/Azure/Azure-Sentinel

This rule makes use of the series decompose anomaly method to detect clients with a high NXDomain response count, which could be indicative of a DGA (cycling through possible C2 domains where most C2s are not live). An alert is generated when new IP address DNS activity is identified as an outlier when compared to the baseline, indicating a recurring pattern. It utilizes ASIM normalization and is applied to any source that supports the ASIM DNS schema.

MITRE ATT&CK coverage

TacticTechniques
Command & Control

Telemetry coverage

ProviderRecord / event type
SysmonEvent ID 22: DNSEvent (DNS query)

Rule body

id: 01191239-274e-43c9-b154-3a042692af06
name: Potential DGA(Domain Generation Algorithm) detected via Repetitive Failures - Anomaly based (ASIM DNS Solution)
description: |
  'This rule makes use of the series decompose anomaly method to detect clients with a high NXDomain response count, which could be indicative of a DGA (cycling through possible C2 domains where most C2s are not live). An alert is generated when new IP address DNS activity is identified as an outlier when compared to the baseline, indicating a recurring pattern. It utilizes [ASIM](https://aka.ms/AboutASIM) normalization and is applied to any source that supports the ASIM DNS schema.'
severity: Medium
status: Available 
tags:
  - Schema: ASimDns
    SchemaVersion: 0.1.6
requiredDataConnectors: []
queryFrequency: 1d
queryPeriod: 14d
triggerOperator: gt
triggerThreshold: 0
tactics:
  - CommandAndControl
relevantTechniques:
  - T1568
  - T1008
query: |
  let threshold = 2.5;
  let min_t = ago(14d);
  let max_t = now();
  let timeframe = 1d;
  // calculate avg. eps(events per second)
  let eps = materialize (_Im_Dns
    | project TimeGenerated
    | where TimeGenerated > ago(5m)
    | count
    | extend Count = Count / 300);
  let maxSummarizedTime = toscalar (
    union isfuzzy=true 
        (
        DNS_Summarized_Logs_ip_CL 
        | where EventTime_t >= min_t
        | summarize max_TimeGenerated=max(EventTime_t)
        | extend max_TimeGenerated = datetime_add('hour', 1, max_TimeGenerated)
        ),
        (
        print(min_t)
        | project max_TimeGenerated = print_0
        )
    | summarize maxTimeGenerated = max(max_TimeGenerated) 
    );
  let summarizationexist = materialize(
    union isfuzzy=true 
        (
        DNS_Summarized_Logs_ip_CL
        | where EventTime_t > ago(1d) 
        | project v = int(2)
        ),
        (
        print int(1) 
        | project v = print_0
        )
    | summarize maxv = max(v)
    | extend sumexist = (maxv > 1)
    );
  let allData = union isfuzzy=true
        (
        (datatable(exists: int, sumexist: bool)[1, false]
        | where toscalar(eps) > 1000
        | join (summarizationexist) on sumexist)
        | join (
            _Im_Dns(responsecodename='NXDOMAIN', starttime=todatetime(ago(2d)), endtime=now())
            | where TimeGenerated > maxSummarizedTime
            | summarize Count=count() by SrcIpAddr, DnsQuery, bin(TimeGenerated, 1h)
            | extend EventTime = TimeGenerated, Count = toint(Count), exists=int(1)
            )
            on exists
        | project-away exists, maxv, sum*
        ),
        (
        (datatable(exists: int, sumexist: bool)[1, false]
        | where toscalar(eps) between (501 .. 1000)
        | join (summarizationexist) on sumexist)
        | join (
            _Im_Dns(responsecodename='NXDOMAIN', starttime=todatetime(ago(3d)), endtime=now())
            | where TimeGenerated > maxSummarizedTime
            | summarize Count=count() by SrcIpAddr, DnsQuery, bin(TimeGenerated, 1h)
            | extend EventTime = TimeGenerated, Count = toint(Count), exists=int(1)
            )
            on exists
        | project-away exists, maxv, sum*
        ),
        (
        (datatable(exists: int, sumexist: bool)[1, false]
        | where toscalar(eps) <= 500
        | join (summarizationexist) on sumexist)
        | join (
            _Im_Dns(responsecodename='NXDOMAIN', starttime=todatetime(ago(4d)), endtime=now())
            | where TimeGenerated > maxSummarizedTime
            | summarize Count=count() by SrcIpAddr, DnsQuery, bin(TimeGenerated, 1h)
            | extend EventTime = TimeGenerated, Count = toint(Count), exists=int(1)
            )
            on exists
        | project-away exists, maxv, sum*
        ),
        (
        DNS_Summarized_Logs_ip_CL
        | where EventTime_t > min_t and EventResultDetails_s == 'NXDOMAIN'
        | project-rename
            SrcIpAddr=SrcIpAddr_s,
            DnsQuery=DnsQuery_s,
            Count=count__d,
            EventTime=EventTime_t
        | extend Count = toint(Count)
        );
  allData
  | make-series QueryCount=dcount(DnsQuery) on EventTime from min_t to max_t step timeframe by SrcIpAddr
  // include calculated Anomalies, Score and Baseline
  | extend (anomalies, score, baseline) = series_decompose_anomalies(QueryCount, threshold, -1, 'linefit')
  | mv-expand anomalies, score, baseline, EventTime, QueryCount
  | extend
    anomalies = toint(anomalies),
    score = toint(score),
    baseline = toint(baseline),
    EventTime = todatetime(EventTime),
    Total = tolong(QueryCount)
  | where EventTime >= ago(timeframe)
  | where score >= threshold * 2
  // Join allData to include DnsQuery details
  | join kind=inner(allData
    | where TimeGenerated >= ago(timeframe)
    | summarize DNSQueries = make_set(DnsQuery, 1000) by SrcIpAddr)
    on SrcIpAddr
  | project-away SrcIpAddr1
entityMappings:
  - entityType: IP
    fieldMappings:
      - identifier: Address
        columnName: SrcIpAddr
eventGroupingSettings:
  aggregationKind: AlertPerResult
customDetails:
  DNSQueries: DNSQueries
  AnomalyScore: score
  baseline: baseline
  Total: Total
alertDetailsOverride:
  alertDisplayNameFormat: "[Anomaly] Potential DGA (Domain Generation Algorithm) originating from client IP: '{{SrcIpAddr}}' has been detected."
  alertDescriptionFormat: "Client has been identified with high NXDomain count which could be indicative of a DGA (cycling through possible C2 domains where most C2s are not live). This client is found to be communicating with multiple Domains which do not exist.\n\nBaseline Domain or DNS query count from this client: '{{baseline}}'\n\nCurrent Domain or DNS query count from this client: '{{Total}}'\n\nDNS queries requested by this client inlcude: '{{DNSQueries}}'"
version: 1.0.2
kind: Scheduled

Stages and Predicates

Stage 0: let

let threshold = 2.5;
let min_t = ago(14d);
let max_t = now();
let timeframe = 1d;
let eps = materialize (_Im_Dns
  | project TimeGenerated
  | where TimeGenerated > ago(5m)
  | count
  | extend Count = Count / 300);
let maxSummarizedTime = toscalar (
  union isfuzzy=true 
      (
      DNS_Summarized_Logs_ip_CL 
      | where EventTime_t >= min_t
      | summarize max_TimeGenerated=max(EventTime_t)
      | extend max_TimeGenerated = datetime_add('hour', 1, max_TimeGenerated)
      ),
      (
      print(min_t)
      | project max_TimeGenerated = print_0
      )
  | summarize maxTimeGenerated = max(max_TimeGenerated) 
  );
let summarizationexist = materialize(
  union isfuzzy=true 
      (
      DNS_Summarized_Logs_ip_CL
      | where EventTime_t > ago(1d) 
      | project v = int(2)
      ),
      (
      print int(1) 
      | project v = print_0
      )
  | summarize maxv = max(v)
  | extend sumexist = (maxv > 1)
  );
let allData = union <inlined as stages below>;

Stage 1: source

let allData

Stage 2: union

union of 4 branches

Stage 3: source

datatable

Stage 4: where

where /* macro: (toscalar(eps) > 1000) */

Stage 5: join

join (summarizationexist) on sumexist

Stage 6: join

join (_Im_Dns) on exists

Stage 7: project-away

project-away exists, maxv, sum*

Stage 8: source

datatable

Stage 9: where

where /* macro: (toscalar(eps) between (501 .. 1000)) */

Stage 10: join

join (summarizationexist) on sumexist

Stage 11: join

join (_Im_Dns) on exists

Stage 12: project-away

project-away exists, maxv, sum*

Stage 13: source

datatable

Stage 14: where

where /* macro: (toscalar(eps) <= 500) */

Stage 15: join

join (summarizationexist) on sumexist

Stage 16: join

join (_Im_Dns) on exists

Stage 17: project-away

project-away exists, maxv, sum*

Stage 18: source

DNS_Summarized_Logs_ip_CL

Stage 19: where

where EventResultDetails_s =~ "NXDOMAIN" and EventTime_t > min_t

Stage 20: project-rename

project-rename

Stage 21: extend

extend Count

Stage 22: summarize

summarize QueryCount by SrcIpAddr

Stage 23: extend

extend anomalies, baseline, score

Stage 24: mv-expand

mv-expand anomalies

Stage 25: extend

extend EventTime, Total, anomalies, baseline, score

Stage 26: where

where EventTime >= ago(86400s)

Stage 27: where

where score >= 5

Stage 28: join

join kind=inner (allData) on SrcIpAddr

Stage 29: project-away

project-away SrcIpAddr1

Indicators

These rows show field, operator, and value matches.

FieldKindValuesSearch
EventResultDetails_seq
  • NXDOMAIN corpus 2 (kusto 2)
field:"EventResultDetails_s" kind:eq value:"NXDOMAIN"
EventTime_tcross_field_compare
  • min_t transforms: op:gt corpus 4 (kusto 4)
field:"EventTime_t" kind:cross_field_compare value:"min_t"
TimeGeneratedcross_field_compare
  • maxSummarizedTime transforms: op:gt corpus 4 (kusto 4)
field:"TimeGenerated" kind:cross_field_compare value:"maxSummarizedTime"
scorege
  • 5 corpus 6 (kusto 6)
field:"score" kind:ge value:"5"

Output fields

These fields are emitted when the rule matches.

FieldSource
QueryCountsummarize
SrcIpAddrsummarize
anomaliesextend
baselineextend
scoreextend
EventTimeextend
Totalextend