Transforming Cmds (5%)
- Transforming: chart timechart stats top rare contingency. NOT: table, fields, eval.
- chart <fn> over A by B (1 over + 1 by; over=x-axis).
- timechart x-axis always _time; span=; only ONE by.
- top=most common (+count+percent, limit=10); rare=least.
- stats fns: count sum avg min max values(uniq mv) list(all) dc first last perc95 stdev.
Filter/Format (10%)
- eval = make/overwrite a field (if, case, coalesce, round, tostring, strftime). Does NOT filter.
- search = match keywords/values (can repeat mid-pipe).
- where = boolean eval expr; only one that compares field-to-field + eval fns (like, cidrmatch, isnull).
- fillnull default replace = 0; fillnull value="N/A".
"A > B (two fields)" → where, not search.
Correlating (15% ★)
transaction
- Groups events → 1 mv event. Opts: maxspan maxpause startswith endswith.
- Auto fields: duration, eventcount.
- Expensive, doesn't scale.
stats
- stats … by id = fast, scalable. Default choice.
Use transaction ONLY for: keep raw text · start/end conditions · time-bound group. Else → stats.
Fields (10%)
- Field Extractor (FX): Regex (unstructured, highlight samples) vs Delimiter (structured, comma/tab/pipe).
- Inline: rex field=_raw "(?<u>\w+)"; erex = example-based.
- Search-time (schema-on-read).
Alias vs Calc Field (10%)
| Alias | Calculated |
| alt name for existing field | new field via eval |
| no new data | computed at search time |
| src_ip⇄clientip | bw = bytes/1024 |
Tags vs Event Types (10%)
| Event type | Tag |
| saved search string | label on field=value |
| → eventtype=failed_login | many tags per value |
| no pipes/time/transforming | normalize across sources |
Macros (10%)
- Settings ▸ Advanced Search ▸ Search macros (macros.conf).
- Call with backticks: `m` / `m(a,b)`.
- Args inside def = $arg$; count in name m(2).
- Optional validation expression.
Workflow Actions (10%)
| Type | Does |
| GET | open URL/link, field vals in query string |
| POST | HTTP POST with field vals |
| Search | launch secondary Splunk search |
Settings ▸ Fields ▸ Workflow actions; scoped to fields/sourcetypes.
Data Models + Pivot (10%)
- Data model = hierarchical semantic map; datasets/objects of type event / search / transaction; attributes (auto, eval, lookup, regex, geo). Children inherit constraints.
- Pivot = point-click UI on top of a data model (no SPL).
- Can't use Pivot without a data model.
- Accelerate → query with tstats.
CIM (10%)
- Splunk_SA_CIM add-on = standard field names + tags + models (Auth, Network, Web, Malware).
- Purpose: normalize at search time so apps (ES) work across sources.
- Make compliant: map fields → CIM names + apply CIM tags.
Search-Time Order
Extraction → Aliases → Calculated fields → Lookups → Event types → Tags
Alias before calc field (calc can use alias); tags applied last.
Top Traps
transaction vs stats (default stats) · alias(rename) vs calc(eval-new) · tag(label) vs eventtype(search) · GET/POST/Search · macro backticks+$arg$ · pivot needs data model · timechart x=_time · where compares fields
65 Q / 57 min ≈ 52s each. Flag & return. Some Q = multi-answer. Avoid dumps.