analyzing-security-logs-with-splunk（使用 splunk 分析安全日志）

Analyzing Security Logs with Splunk

When to Use

• Investigating a security incident that requires correlation across multiple log sources
• Hunting for adversary activity using known TTPs and IOCs
• Building detection rules for specific attack patterns
• Reconstructing an incident timeline from disparate log sources
• Analyzing authentication anomalies, lateral movement, or data exfiltration patterns

Do not use for real-time packet-level analysis; use Wireshark or Zeek for full packet capture analysis.

Prerequisites

• Splunk Enterprise or Splunk Cloud with Enterprise Security (ES) app installed
• Log sources ingested: Windows Event Logs (via Splunk Universal Forwarder or WEF), firewall, proxy, DNS, EDR, email gateway
• Splunk CIM (Common Information Model) data models configured for normalized field names
• SPL proficiency at intermediate level or higher
• Role-based access with search and accelerate_search capabilities in Splunk

Workflow

Step 1: Scope the Investigation in Splunk

Define search parameters based on incident triage data:

| Set initial investigation scope
index=windows OR index=firewall OR index=proxy
  earliest="2025-11-14T00:00:00" latest="2025-11-16T00:00:00"
  (host="WKSTN-042" OR src_ip="10.1.5.42" OR user="jsmith")
| stats count by index, sourcetype, host
| sort -count

This query establishes which log sources contain relevant data for the investigation timeframe and affected assets.

Step 2: Analyze Authentication Events

Investigate suspicious authentication patterns using Windows Security Event Logs:

| Detect brute force and credential stuffing
index=windows sourcetype="WinEventLog:Security" EventCode=4625
  earliest=-24h
| stats count as failed_attempts, values(src_ip) as source_ips,
  dc(src_ip) as unique_sources by TargetUserName
| where failed_attempts > 10
| sort -failed_attempts

| Detect pass-the-hash (Logon Type 9 - NewCredentials)
index=windows sourcetype="WinEventLog:Security" EventCode=4624
  Logon_Type=9
| table _time, host, TargetUserName, src_ip, LogonProcessName

| Detect lateral movement via RDP
index=windows sourcetype="WinEventLog:Security" EventCode=4624
  Logon_Type=10
| stats count, values(host) as targets by TargetUserName, src_ip
| where count > 3
| sort -count

Step 3: Trace Process Execution

Use Sysmon logs to reconstruct process execution chains:

| Process creation with parent chain (Sysmon Event ID 1)
index=sysmon EventCode=1 host="WKSTN-042"
  earliest="2025-11-15T14:00:00" latest="2025-11-15T15:00:00"
| table _time, ParentImage, ParentCommandLine, Image, CommandLine, User, Hashes
| sort _time

| Detect suspicious PowerShell execution
index=sysmon EventCode=1 Image="*\\powershell.exe"
  (CommandLine="*-enc*" OR CommandLine="*-encodedcommand*"
   OR CommandLine="*downloadstring*" OR CommandLine="*iex*")
| table _time, host, User, ParentImage, CommandLine
| sort _time

| Detect LSASS credential dumping
index=sysmon EventCode=10 TargetImage="*\\lsass.exe"
  GrantedAccess=0x1010
| table _time, host, SourceImage, SourceUser, GrantedAccess

Step 4: Analyze Network Activity

Correlate network logs with endpoint events:

| Detect C2 beaconing pattern
index=proxy OR index=firewall dest_ip="185.220.101.42"
| timechart span=1m count by src_ip
| where count > 0

| Detect DNS tunneling (high query volume to single domain)
index=dns
| rex field=query "(?<subdomain>[^\.]+)\.(?<domain>[^\.]+\.[^\.]+)$"
| stats count, avg(len(query)) as avg_query_len by domain, src_ip
| where count > 500 AND avg_query_len > 40
| sort -count

| Detect large data transfers (potential exfiltration)
index=proxy action=allowed
| stats sum(bytes_out) as total_bytes by src_ip, dest_ip, dest_host
| eval total_MB=round(total_bytes/1024/1024,2)
| where total_MB > 100
| sort -total_MB

Step 5: Build the Incident Timeline

Reconstruct a unified timeline across all log sources:

| Unified incident timeline
index=windows OR index=sysmon OR index=proxy OR index=firewall
  (host="WKSTN-042" OR src_ip="10.1.5.42" OR user="jsmith")
  earliest="2025-11-15T14:00:00" latest="2025-11-15T16:00:00"
| eval event_summary=case(
    sourcetype=="WinEventLog:Security" AND EventCode==4624, "Logon: ".TargetUserName." from ".src_ip,
    sourcetype=="WinEventLog:Security" AND EventCode==4625, "Failed logon: ".TargetUserName,
    sourcetype=="XmlWinEventLog:Microsoft-Windows-Sysmon/Operational" AND EventCode==1,
      "Process: ".Image." by ".User,
    sourcetype=="proxy", "Web: ".http_method." ".url,
    1==1, sourcetype.": ".EventCode)
| table _time, sourcetype, host, event_summary
| sort _time

Step 6: Create Detection Rules

Convert investigation findings into persistent Splunk correlation searches:

| Correlation search: PowerShell spawned by Office applications
index=sysmon EventCode=1
  Image="*\\powershell.exe"
  (ParentImage="*\\winword.exe" OR ParentImage="*\\excel.exe"
   OR ParentImage="*\\outlook.exe")
| eval severity="high"
| eval mitre_technique="T1059.001"
| collect index=notable_events

Key Concepts

Tools & Systems

• Splunk Enterprise Security (ES): Premium SIEM application providing correlation searches, risk-based alerting, and investigation workbench
• Splunk SOAR: Orchestration platform integrated with Splunk ES for automated response playbooks
• Sysmon: Microsoft system monitoring tool providing detailed process, network, and file change telemetry ingested into Splunk
• Splunk Attack Analyzer: Automated threat analysis that detonates suspicious files and URLs, feeding results into Splunk
• BOSS of the SOC (BOTS): SANS/Splunk training dataset for practicing incident investigation SPL queries

Common Scenarios

Scenario: Investigating Credential Stuffing Leading to Account Takeover

Context: Security operations receives an alert for multiple successful logins to a single account from geographically dispersed IP addresses within a 30-minute window.

Approach:

1. Query Event ID 4624 for the affected account to map all login sources and times
2. Correlate login IPs against threat intelligence feeds using a Splunk lookup table
3. Check proxy logs for suspicious activity from the authenticated sessions
4. Search for lateral movement from the compromised account (Event ID 4624 Type 3 to other hosts)
5. Build a timeline showing credential stuffing attempts, successful login, and post-compromise activity
6. Create a correlation search to detect similar patterns on other accounts

Pitfalls:

• Searching only the last 24 hours when the credential stuffing may have occurred over weeks
• Not checking for VPN logs that may show the same account authenticating from impossible travel distances
• Failing to normalize timestamps across log sources in different time zones

Output Format

SPLUNK INVESTIGATION REPORT
============================
Incident:        INC-2025-1547
Analyst:         [Name]
Investigation Period: 2025-11-14 00:00 UTC - 2025-11-16 00:00 UTC

SEARCH SCOPE
Indexes:         windows, sysmon, proxy, firewall, dns
Hosts:           WKSTN-042, SRV-FILE01
Users:           jsmith, svc-backup
Source IPs:      10.1.5.42, 10.1.10.15

KEY FINDINGS
1. [timestamp] - Initial compromise via phishing (Sysmon Event 1)
2. [timestamp] - C2 established (proxy logs, beacon pattern detected)
3. [timestamp] - Credential theft (Sysmon Event 10, LSASS access)
4. [timestamp] - Lateral movement to SRV-FILE01 (Event 4624 Type 3)
5. [timestamp] - Data staging and exfiltration (proxy bytes_out anomaly)

SPL QUERIES USED
[numbered list of key queries with descriptions]

DETECTION GAPS IDENTIFIED
- No Sysmon deployed on SRV-FILE01 (blind spot)
- Proxy logs missing SSL inspection for C2 domain
- PowerShell ScriptBlock logging not enabled

RECOMMENDED DETECTIONS
1. Correlation search for Office-spawned PowerShell
2. Threshold alert for LSASS access patterns
3. Behavioral rule for beacon-interval network traffic