Migrating from UDM
Map retired UDM field names to OCSF 1.9 paths when you port saved searches, detection rules, dashboards and parsers.
Map retired UDM names to OCSF 1.9 paths when you port saved searches, detection rules, dashboard panels, notebook queries, and parsers.
Older nano installs used the Unified Data Model (UDM), with flat columns such as src_ip, process_name, and file_hash and source-specific extras in an ext JSON column. New installs store and query OCSF in the ClickHouse table ocsf_logs; see OCSF schema. The UDM profile (NANO_SCHEMA_PROFILE=udm) can still be chosen at install, and an existing install keeps the profile it was installed with, but UDM is being retired.
Old forms on the OCSF profile
| Old form | What happens | Use instead |
|---|---|---|
ext.foo | Resolves to unmapped.foo | unmapped.foo |
source_type | Resolves to the log source routing key | metadata.log_source |
file_hash, process_hash | Resolve to the derived SHA-256 columns | file.hashes.value / process.file.hashes.value (every algorithm), or file.hashes.sha256 |
src_ip, user, src_host and other UDM names | Resolve to their OCSF column (class-split names such as user read a unified column) while the compatibility window lasts | The OCSF path from the tables below |
file.hashes.sha256, process.file.hashes.sha256 and actor.process.file.hashes.sha256 are nano-derived columns: the SHA-256 element of each hashes[] list, lowercased.
Field map
Each UDM name maps to one canonical OCSF 1.9 path. Where OCSF splits a concept by event class, both paths are listed; see Paths depend on the class.
Event identity
| UDM | OCSF 1.9 | Notes |
|---|---|---|
timestamp | time | |
id | metadata.uid | |
source_type | metadata.log_source | |
message | message | Unchanged |
_raw | raw_data | |
event_bytes | raw_data_size | |
severity | severity_id (number) or severity (label) | Values change, see below |
event_type, action | class_uid + activity_id | Values change, see below |
ext.<key> | unmapped.<key> | Keys are flattened to dotted snake_case |
tags | unmapped.tags |
Network
| UDM | OCSF 1.9 | Notes |
|---|---|---|
src_ip | src_endpoint.ip | |
dest_ip | dst_endpoint.ip | dst_, not dest_ |
src_port | src_endpoint.port | |
dest_port | dst_endpoint.port | |
src_host | src_endpoint.hostname (network classes), device.hostname (endpoint classes) | Class-dependent |
dest_host | dst_endpoint.hostname | |
src_mac, dest_mac | src_endpoint.mac, dst_endpoint.mac | |
protocol | connection_info.protocol_num (6) | IANA number. connection_info.protocol_name (tcp) exists too, but many sources fill only the number |
bytes_in, bytes_out | traffic.bytes_in, traffic.bytes_out | OCSF: bytes_in is destination→source, bytes_out is source→destination |
packets_in, packets_out | traffic.packets_in, traffic.packets_out | |
bytes | traffic.bytes | |
response_time | duration |
Hosts and users
| UDM | OCSF 1.9 | Notes |
|---|---|---|
host, hostname | device.hostname | |
user | user.name or actor.user.name | Class-dependent: the account acted on vs the account acting |
src_user | actor.user.name | The initiating account |
user_domain | user.domain | |
user_id | user.uid | |
email | user.email_addr | |
display_name | user.display_name | |
auth_type | auth_protocol_id (number) or auth_protocol (label) | |
auth_result | status_id (1 Success, 2 Failure) or status (label) | |
mfa_used | is_mfa | |
session_id | session.uid | |
user_identity_* | unmapped.user_identity_* | Directory enrichment fields |
Process
| UDM | OCSF 1.9 | Notes |
|---|---|---|
process_name | process.name (Process Activity) or actor.process.name (other classes) | Class-dependent |
command_line | process.cmd_line or actor.process.cmd_line | Class-dependent |
process_id | process.pid or actor.process.pid | |
process_guid | process.uid or actor.process.uid | |
process_path | process.file.path or actor.process.file.path | |
process_hash | process.file.hashes.value or actor.process.file.hashes.value | |
parent_process_name | actor.process.name | On Process Activity the parent is the actor |
parent_command_line | actor.process.cmd_line | |
parent_process_path | actor.process.file.path |
File
| UDM | OCSF 1.9 | Notes |
|---|---|---|
file_path | file.path | |
file_name | file.name | |
file_hash | file.hashes.value | Never file.hashes.sha256 |
file_action | activity_id (number) or activity_name (label) | Activity numbers depend on the class |
Web, DNS and email
| UDM | OCSF 1.9 | Notes |
|---|---|---|
url | http_request.url.url_string (HTTP) or url.url_string (other classes) | Class-dependent |
url_domain | http_request.url.hostname or url.hostname | |
uri_path | http_request.url.path | |
http_method, method | http_request.http_method | |
http_status_code, http_status | http_response.code | |
status_code | http_response.code for HTTP status | On OCSF, a bare status_code is the event's own status code, not the HTTP code |
status (as HTTP status) | http_response.code | On OCSF, status is the event outcome label |
http_user_agent, user_agent | http_request.user_agent | |
http_referrer, referrer | http_request.referrer | |
query, dns_query | query.hostname | |
answer | answers.rdata | A list |
sender | email.from | |
recipient | email.to | A list |
subject | email.subject | |
message_id | email.message_uid |
Cloud
| UDM | OCSF 1.9 |
|---|---|
cloud_provider | cloud.provider |
cloud_region | cloud.region |
cloud_account_id | cloud.account.uid |
cloud_account_name | cloud.account.name |
cloud_service | api.service.name |
resource_id, resource_name, resource_type | resources.uid, resources.name, resources.type (lists) |
cve | vulnerabilities.cve.uid (a list) |
error_code, error_message | unmapped.error_code, unmapped.error_message |
Enrichment, risk and prevalence
UDM had one column per enrichment. On the OCSF profile, geolocation fills standard endpoint attributes, and the other enrichments are nano-derived columns written at ingest.
| UDM | OCSF profile |
|---|---|
enriched_src_country_code, enriched_dest_country_code | src_endpoint.location.country, dst_endpoint.location.country |
enriched_src_asn, enriched_src_as_name | src_endpoint.autonomous_system.number, src_endpoint.autonomous_system.name (and dst_endpoint equivalents) |
ioc_src_ip_threat_type, ioc_dest_ip_threat_type, ioc_domain_threat_type, ioc_hash_threat_type | enrichments.ioc_src_ip_threat_type, enrichments.ioc_dest_ip_threat_type, enrichments.ioc_domain_threat_type, enrichments.ioc_hash_threat_type |
src_is_tor, dest_is_tor | enrichments.ioc_src_ip_threat_type="tor_exit", enrichments.ioc_dest_ip_threat_type="tor_exit" |
custom_src_ip_tags, custom_dest_ip_tags | enrichments.custom_src_ip_tags, enrichments.custom_dest_ip_tags |
prevalence_* | prevalence_file_hash, prevalence_process_hash, prevalence_dest_domain, prevalence_dest_ip (same names) |
risk_score | risk_score (also risk_level) on Detection Findings, owned by risk-based alerting |
Per-feed details: Enrichment field reference.
Values change too
For some filters the field name and the values both change. A find-and-replace on the name alone gives a query that matches nothing.
| UDM filter | OCSF 1.9 filter | What changed |
|---|---|---|
source_type=sysmon | metadata.log_source=windows_sysmon, or by kind: class_uid=1007 | Use the log source slug, or a numeric class |
event_type=login | class_uid=3002 activity_id=1 | String event type → class + activity numbers |
auth_result=failure | status_id=2 | String → number |
action="created" (file) | class_uid=1001 activity_id=1 | String → class + activity |
severity=high | severity_id=4 | String → normalized number (0–6) |
protocol=tcp | connection_info.protocol_num=6 | Name → IANA number (6 TCP, 17 UDP). Many sources leave connection_info.protocol_name empty |
Where OCSF has both a number and a label (status_id / status, auth_protocol_id / auth_protocol, activity_id / activity_name), filter on the number. Label filters such as status="failure" also work and are case-insensitive. Don't rely on the old names: during the compatibility window some map to number paths (for example auth_type to auth_protocol_id), so an old string value can match nothing.
Before and after
Failed logons by user, legacy form:
event_type=login auth_result=failure | stats count by user, src_ipOCSF 1.9:
class_uid=3002 activity_id=1 status_id=2 | stats count by user.name, src_endpoint.ipEncoded PowerShell launched by Office, legacy form:
process_name="powershell.exe" parent_process_name="winword.exe" command_line="*-enc*"OCSF 1.9:
class_uid=1007 process.name="powershell.exe" actor.process.name="winword.exe" process.cmd_line="*-enc*"Known-bad hash anywhere, legacy form:
file_hash="44d88612fea8a8f36de82e1278abb02f" OR process_hash="44d88612fea8a8f36de82e1278abb02f"OCSF 1.9:
file.hashes.value="44d88612fea8a8f36de82e1278abb02f" OR process.file.hashes.value="44d88612fea8a8f36de82e1278abb02f" OR actor.process.file.hashes.value="44d88612fea8a8f36de82e1278abb02f"A source-specific extra field, legacy form:
source_type=json_generic ext.tenant_tier="gold"OCSF 1.9:
metadata.log_source=json_generic unmapped.tenant_tier="gold"Parsers
UDM parsers wrote .udm.* fields. OCSF parsers build the whole OCSF event on the event root, set class_uid, activity_id, type_uid and metadata.*, keep the original line in raw_data, and let unmapped source fields spill to unmapped. A parser must not write .udm.* or a top-level source_type. See Parser extensions.
UDM-profile installs
An install on the UDM profile keeps its logs table and flat field names; the profile is fixed at install and isn't converted in place. To move to OCSF, install a new deployment on the OCSF profile and point your senders at it, then port saved searches and rules with the tables above. See Deployment models.