Normalization

Normalization is a critical stage connecting ingestion from sources and forwarding to targets used to coalesce log data from diverse sources into consistent formats, enabling unified handling across different logging systems.

Log Formats

The processor supports several widely-used log formats:

Generic

Format	Notation	Key Identifier	Layout Characteristics	Example Fields
Elastic Common Schema (ECS)	Dot notation with lowercase	@timestamp	Hierarchical structure	source.ip, network.direction
Splunk Common Information Model (CIM)	Underscore with lowercase	_time	Flat structure	src_ip, network_direction
Advanced Security Information Model (ASIM)	PascalCase	TimeGenerated	Explicit names	SourceIp, NetworkDirection

Security-specific

Format	Description	Key Identifier	Example Fields
Common Event Format (CEF)	ArcSight's standard format	rt (receiptTime)	networkUser, sourceAddress
Log Event Extended Format (LEEF)	IBM QRadar's format	devTime	networkUser, srcAddr
Common Security Log (CSL)	Microsoft Sentinel's format	TimeGenerated	NetworkUser, SourceAddress

Format Detection

Source formats can be automatically detected using certain characteristic fields, e.g.:

Context	Field	Format
Timestamp	@timestamp	ECS
	_time	CIM
	TimeGenerated	ASIM/CSL
Security	rt	CEF
	devTime	LEEF
CSL detection	TimeGenerated + LogSeverity	CSL
	TimeGenerated only	ASIM

Conversion

Casing and Delimiters

Each format follows specific naming conventions:

ECS	source.ip, event.severity
CIM	src_ip, event_severity
ASIM	SourceIp, EventSeverity
CEF	sourceAddress, eventSeverity
LEEF	srcAddr, evtSev
CSL	SourceIP, EventSeverity

warning

Complex format conversions may impact performance.

Field Mapping

There are identifiable common network fields based on context across various formats:

	Context
Format	Source IP	Destination IP	Direction
ECS	source.ip	destination.ip	network.direction
CIM	src_ip	dest_ip	network_direction
ASIM	SourceIp	DstIp	NetworkDirection
CEF	src	dst	networkDirection
LEEF	srcAddr	dstAddr	netDir
CSL	SourceIp	DestinationIp	NetworkDirection

Configuration

Basic

Convert from ECS to ASIM format:

normalize:
  - source_format: ecs
  - target_format: asim

Field-specific

Convert a specific network field:

normalize:
  - field: network_data
  - source_format: cef
  - target_format: ecs

Auto-detection

Let the processor detect the source format:

normalize:
  - target_format: cim

Best Practices

For data integrity, always validate transformed logs against originals, keep original fields when possible for debugging, and document format-specific transformations.

For performance, do the normalization early in the pipeline, cache results for lookup when possible, and monitor transformation overhead.

For error handling, use ignore_failure and implement fallback mechanisms. Also, do not forget to test with diverse samples.