UEBA Configuration
This topic provides the high-level tasks required to configure UEBA.
Changing the UEBA start-date or the UEBA processed schemas requires a re-run of the UEBA system as well as cleanup of the UEBA databases. In order to avoid deleting the information in the UI, you can use the reset_presidio.py script as described in reset-presidio script, it will keep the data in the UI (e.g. Alerts, Indicators, Entities and Scores).
To configure a single UEBA server, see "Task 3. Install and Configure NetWitness UEBA" under Installation Tasks topic in the UEBA Standalone Installation Guide.
Configure Multiple UEBA Servers
NetWitness Platform now supports installing multiple servers of UEBA in your environment.
Before using this feature, ensure that you meet the following requirements:
The multiple UEBA deployments are independent, with one supporting Logs & Endpoint models and the other dedicated exclusively to Network (TLS) models. Customers with a different use case will need to contact the NetWitness Customer Support team.
Multi-UEBA can use used for the following scenario:
NetWitness recommends that you configure Authentication, File, Active Directory, Process, and Registry schemas on one UEBA server and TLS schema on another UEBA server for better data processing.
Prerequisites
Ensure that the NetWitness Platform and Hosts (UEBA) are in version 12.3 or later.
Procedure
-
Follow the install instructions for installing multiple UEBA servers. For example, UEBA-Server-1 and UEBA-Server-2. For more information, see "Task 3. Install and Configure NetWitness UEBA" under Installation Tasks topic in the UEBA Standalone Installation Guide.
You can configure multiple UEBA servers in your environment. NetWitness has installed and verified up to three UEBA servers.
-
Follow the ueba-server-config script to set up data schemas on the installed UEBA server 1 and UEBA server 2. For more information, see ueba-server-config script.
Best Practices to Add and Remove Schemas for Multiple UEBA Servers
If you are planning to install multiple UEBA servers in your environment. Consider that you have all six schemas configured in the 12.2 or an earlier version of the UEBA server.
NetWitness recommends that the TLS schema (Network data) must be configured on the new UEBA Server first, and then the existing UEBA server containing all schemas must be reset and re-configured with the five schemas Authentication, File, Active Directory, Process, and Registry (Logs and Endpoint data). For more information on configuration, see ueba-server-config script. You need to reset the start date as well and ensure you set the start date one month back from the current date. For more information, see reset-presidio script.
ueba-server-config script
The ueba-server-config script is usually used to configure and run the UEBA component after the deployment. Also, it can be used to update the UEBA configuration during run time.
If you change the start-time or the processing schemas, you must re-run UEBA. All script arguments (except the boolean arguments) are mandatory and must be filled.
For more information on the script parameters, see the NetWitness Standalone Installation Guide for Version 12.4.
To run the script use the following command /opt/rsa/saTools/bin/ueba-server-config --help
-u |
<user> |
User name of the credentials for the Broker or Concentrator instance that you are using as a data source.
|
-p |
<password> |
Password of the credentials for the Broker or Concentrator instance that you are using as a data source. The following special characters are supported in a password.
!"#$%&()*+,-:;<=>?@[\]^_`\{|}
If you want to include a special character or special characters, you must delimit the password with an apostrophe sign, for example:
sh /opt/rsa/saTools/bin/ueba-server-config -u brokeruser -p '!"UHfz?@ExMn#$' -h 10.64.153.104 -t 2018-08-01T00:00:00Z -s 'AUTHENTICATION FILE ACTIVE_DIRECTORY TLS PROCESS REGISTRY' -o broker -v
|
-h |
<host> |
IP address of the Broker or Concentrator used as the data source. Currently, only one data source is supported.
|
-o |
<type> |
Data source host type (broker or concentrator).
|
-t |
<startTime> |
Historical start time as of which you start collecting data from the data source in YYYY-MM-DDTHH-MM-SSZ format (for example, 2018-08-15T00:00:00Z).
The script interprets the time you enter as UTC (Coordinated Universal Time) and it does not adjust the time to your local time zone.
|
-s |
<schemas> |
Array of data schemas. If you want to specify multiple schemas, use a space to separate each schema (for example, AUTHENTICATION FILE ACTIVE_DIRECTORY PROCESS REGISTRY TLS).
|
-v |
|
verbose mode.
|
-e |
<argument> |
Boolean Argument. This enables the UEBA indicator forwarder to Respond.
If your NetWitness deployment includes an active Respond server, you can transfer NetWitness UEBA indicators to the Respond server and create incidents by enabling the indicator forwarder, from this data. For more information on how to enable the NetWitness UEBA incidents aggregation, see Enable User Entity Behavior Analytics Incident Rule.
|
The TLS packet requires adding the hunting package. For more information, see Add Features for UEBA Packet Schema.
reset-presidio script
The reset_presidio.py script deletes the UEBA back-end databases and can also delete the front-end database that is present in the UI.
The reset_presidio.py script is used to re-run the UEBA system as well as to update the UEBA start-date and the processing schemas easily without having to provide all the other parameters required by the ueba-server-config script. This script re-runs the UEBA while it deletes the backed data (models, aggregations, etc.). To delete the front-end data (UI entities and alerts, etc.) use the clean option. If you don’t specify a date, the script will set the default start date, a 28 days earlier than the current date. recommends that the UEBA start date is set to 28 days earlier than the current date. For UEBA systems that intend to process TLS data, you must verify that the start date is set to no later than 14 days earlier than the current date.
UEBA requires to process 28 days of data before the alerts can be created.
• If you choose a start date that is less than 28 days before the current date, for example 10 days earlier from the current date, you will have to wait for another 18 days from the current date to see alerts in your UEBA system (if created).
• If you choose a start date that is greater than 27 days, it's recommended to delete the front-end database as well (use the -c) to avoid duplicate alerts.
To run the script, load the Airflow virtual environment variables as follows:
-
source /etc/sysconfig/airflow
-
source $AIRFLOW_VENV/bin/activate
-
python /var/netwitness/presidio/airflow/venv39/lib/python3.9/site-packages/presidio_workflows-1.0-py3.9.egg/presidio/utils/airflow/reset_presidio.py --help
-
deactivate
-h, --help |
|
Script Help |
-c, --clean |
<argument> |
Clean any existing data in Elasticsearch DB (as Alerts, Indicators, Entities, etc), all data will be deleted form the UEBA UI |
-s |
<schema> |
Reconfigure the UEBA engine array of schemas (e.g. [AUTHENTICATION FILE ACTIVE_DIRECTORY PROCESS REGISTRY TLS]) |
-d |
<date> |
Reconfigure the UEBA engine to start from midnight UTC of this date. If not set, by default reset the start date to 27 days before the current system day, at midnight UTC, to avoid duplicate alerts in the UEBA UI, in case you didn't cleaned the elasticsearch data (-c) (e.g. 2010-12-31) |
Please refer to the above table for the required arguments to pass along with the reset command. For more information, refer to the example command below.
python /var/netwitness/presidio/airflow/venv39/lib/python3.9/site-packages/presidio_workflows-1.0-py3.9.egg/presidio/utils/airflow/reset_presidio.py -c -d 2023-11-16 -s AUTHENTICATION ACTIVE_DIRECTORY FILE PROCESS REGISTRY TLS
Add a Schema without Rerunning the UEBA
Adding a schema without rerunning the UEBA system is supported on NetWitness Platform.
To add a new UEBA schema without rerunning the UEBA system, run the following command on the UEBA host.
curl -X PATCH http://localhost:8881/configuration -H 'content-type: application/json' -d '{"operations":[{"op":"add","path":"/dataPipeline/schemas/-","value":"<SCHEMA>"}]}'
Where <SCHEMA> string can be replaced with any one of the following schemas:
- AUTHENTICATION
- FILE
- ACTIVE_DIRECTORY
- PROCESS
- REGISTRY
- TLS
UEBA Indicator Forwarder
The UEBA Indicator Forwarder is supported by the NetWitness UEBA.
If your NetWitness environment includes an active respond server, you can transfer the UEBA indicators to the respond server and to the correlation server in order to create Incidents. For more information, see Enable User Entity Behavior Analytics Incident Rule.
Run the following command to activate the UEBA Indicator Forwarder:
curl -X PATCH http://localhost:8881/configuration -H 'content-type: application/json' -d '{"operations":[{"op":"replace","path":"/outputForwarding/enableForwarding","value":true}]}'
To deactivate the UEBA indicator forwarder, change the “value":true at the request body to be “value":false.
Update Data Source Details
In order to update the details of the data source you must use the ueba-server-config script. For more information, see ueba-server-config script.
From 12.3 version or later, if you change the data source using the ueba-server-config script, UEBA will use the previously configured start date by default. To change the UEBA start date, use the reset-presidio script.
The data sources details are:
- Data Source type (Broker / Concentrator).
- Data Source username.
- Data Source password.
- Data Source host.
Add Features for UEBA Packet Schema
Add the Hunting Pack:
In NetWitness Platform, add the hunting pack or verify it it’s available:
- Log in to the NetWitness Platform.
- Navigate to (Admin) and select Admin Server.
- Click and select Configure > Live Content.
- On the left menu, select the following:
- Bundle under Resources Type.
- Packet under Medium
- Click Search.
A list of matching resources is displayed.
- Select Hunting Pack from the list and click Deploy.
The hunting pack is added.
Add JA4
The JA4 fields are supported by the Packet Decoder in 12.5 and later. Verify that your Packet Decoder is upgraded to one of these versions.
To add JA4
1. Log in to the NetWitness Platform.
2. Go to (Admin) > Services and select Decoder.
3. Navigate to /decoder/parsers/config/parsers.options.
4. Add HTTPS="ja4=true.
Assign User Access to UEBA
To create a user with privileges to access the UEBA pages (Users tab) on the Netwitness UI do the following:
- Navigate to (Admin) > Security.
- Create a new UEBA_Analysts and Analysts user roles.
For more information, see the "Manage Users with Roles and Permissions" topic in the System Security and User Management Guide.
Create an Analysts Role
In order to fetch data from the data source (Broker / Concentrator), you need to create a user using the Analysts role in the data source service.
- Navigate to the security tab at the data source service page.
- Go to the (Admin) > Services > Security.
- Create an analyst user and assign it to the any of supported special characters.
Enable User Entity Behavior Analytics Incident Rule
In order to aggregate the UEBA indicators under Incident rule, follow the instructions below:
Enable the UEBA Forwarding process as described in Enable UEBA Indicator Forwarder.
By default, the NetWitness UEBA (On-premises) rules are disabled in your environment. You can enable them to generate the incident IDs for the alerts and customize the NetWitness UEBA (On-premises) rules settings.
- Go to > Incident Rules.
- Select the NetWitness UEBA (On-premises) rule and click Enable.
A confirmation pop-up is displayed.
- Click OK.
Enable or Disable Modeled Behaviors for Users
The UEBA Modeled Behaviors functionality is enabled by default.
To disable perform the following:
-
SSH to the UEBA server.
Edit and add the parameter entity.profile.enabled=false in the file/etc/netwitness/presidio/configserver/configurations/presidio-uiconf.properties .
-
Run the command to restart the presidio-ui.
systemctl restart presidio-ui
To enable, remove the parameter entity.profile.enabled=false from the file and restart the presidio-ui using the step 2.
Once you have enabled or disabled the Modeled Behaviors, you can verify from NetWitness Platform UI.
To verify, perform the following:
- Log in to the NetWitness Platform and click Users.
- In the Overview tab, under Top Risky Users panel, click on a username.
- Click the Modeled Behaviors tab.
For more information, see "View Modeled Behaviors" topic in the UEBA User Guide for NetWitness Platform 12.3.
Update UEBA Queries using nw-shell Utility
This topic describes the steps required to update the queries for UEBA schemas using nw-shell Utility.
Prior to version 12.5, queries for schemas were stored and accessed from the schemas.json file. Starting with version 12.5, these queries have been migrated to the UEBA server database instead of the schemas.json file. If the queries for schemas were modified in the schemas.json file before version 12.5, they will be transferred to the UEBA server database as part of the post-upgrade process.
Users can update queries for schema in UEBA server database using nw-shell utility.
Note:
- If users have added new meta fields using custom parsers, those meta fields will not be included in the default queries defined for the schema. Users must update the query using the nw-shell utility.
- Only administrators can update schema queries using specific commands within the nw-shell utility.
To Update UEBA Queries using nw-shell Utility
-
SSH into the Admin server.
-
Enter the following command: nw-shell
The console window is displayed.
-
Connect to ueba-server using the following command:
connect --service ueba-server
-
Enter the login command: login
-
Enter the admin username and password.
-
Navigate to ueba/mapping using the following command: cd ueba/mapping
-
Run the following command to list available options/commands: ls
-
Navigate to the following available commands and invoke them. cd <command>
Refer to the following table for the command and its usage.
get-mapping |
Get details/mappings of all schemas |
cd get-mapping invoke
|
update-mapping
|
Updating details/queries/mapping of schema |
cd update-mapping invoke --file <json-input-file-path>
Example of the Json file:
{
"id":"662a21d8dfcc3b23d78b25d7",
"selectClause":"select ueba.schema,sessionid,event.source.id,agent.id,event.time,device.type,user.src,action,alias.host,owner,directory.src,filename.src,ec.subject,registry.key,cert.common,dir.path.src,nwe.callback_id,file.cat.src ",
"query":"where category='Registry Event' AND device.type='nwendpoint'",
"metaMapping":{
"time":"event.time"
},
"multiValue":"action,alias.host,dir.path.src,file.cat.src"
}
Note: You can modify only the values of the following fields: selectClause, query, metaMapping, and multiValue.
|
Removal of Packetbeat Service
From the 12.3 version or later, the Packbeat service has been removed from UEBA to improve memory usage and performance. This allows other services in UEBA to utilize the resources more efficiently, reducing the load on the system.
Learning Period Per Scale
Learning Period Per Scale for 12.5
Note: The Learning Period Per Scale numbers for the Physical Machine, Virtual Machine, and Multi-UEBA deployments remain the same in versions 12.5 and 12.5.1.
Physical Machine (SERIES 6 ESA (DELL R640) SPECIFICATIONS)
150,000 users with 50 million Log and Endpoint events + 70 million Network events with 150,000 JA4 entities per day
|
Yes |
Fresh Installation of 12.5
Up to 8 days with 28 days of historical data
|
150,000 users with 50 million Log and Endpoint events + 70 million Network events with 150,000 JA4 entities per day
|
Yes |
Upgrade from 12.2.x, 12.3, 12.3.1, 12.4, 12.4.1, and 12.4.2 to 12.5
UEBA reset is not required.
|
150,000 users with 50 million Log and Endpoint events + 70 million Network events with 150,000 JA4 entities per day
|
Yes |
Upgrade from 12.2.x, 12.3, 12.3.1, 12.4, 12.4.1, and 12.4.2 to 12.5 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required.
UEBA reset is required.
Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
150,000 users with 50 million Log and Endpoint events + 70 million Network events with 150,000 JA4 entities per day
|
No |
Fresh installation of 12.5
28 days to complete the learning period before generating alerts.
|
Virtual Machine
The recommended vCPU specification for UEBA is Intel Xeon CPU @2.59 Ghz.
32 cores |
256GB |
192GB
|
-
Storage: 1.5 TB
-
Provisioning: Thick
|
500 |
500
|
To determine the scale limits for Virtual Machine deployments, refer to the Scaling Limitation Issue section in the Troubleshooting UEBA Configurations.
IMPORTANT:
- You must reserve all the resources allocated to UEBA on the VM server. For example, if a user has a 2.1GHz CPU, then 32CPUs * 2.1GHz = 67.2GHz or 67200MHz must be reserved.
- The /var/netwitness partition must be mounted on a 1.5 TB Thick-provisioned disk for storage usage.
Note: NetWitness recommends you to deploy UEBA on a virtual host, only if your log collection volume is low. If you have a moderate to high log collection volume, NetWitness recommends you to deploy UEBA on the physical host as described in the "NetWitness UEBA Host Hardware Specifications" topic of the Physical Host Installation Guide. Contact NetWitness Customer Support (https://community.netwitness.com/t5/support-information/how-to-contact-netwitness-support/ta-p/563897) for advice on choosing which host, virtual or physical, to use for UEBA.
150,000 users with 50 million Log and Endpoint events + 25 million Network events with 150,000 JA4 entities per day
|
Yes |
Fresh Installation of 12.5
Up to 8 days with 28 days of historical data
|
150,000 users with 50 million Log and Endpoint events + 25 million Network events with 150,000 JA4 entities per day
|
Yes |
Upgrade from 12.2.x, 12.3, 12.3.1, 12.4, 12.4.1, and 12.4.2 to 12.5
UEBA reset is not required.
|
150,000 users with 50 million Log and Endpoint events + 25 million Network events with 150,000 JA4 entities per day
|
Yes |
Upgrade from 12.2.x, 12.3, 12.3.1, 12.4, 12.4.1, and 12.4.2 to 12.5 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required.
Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
150,000 users with 50 million Log and Endpoint events + 25 million Network events with 150,000 JA4 entities per day
|
No |
Fresh installation of 12.5
28 days to complete the learning period before generating alerts.
|
Note: Network events per day refers to number of events consumed by UEBA per day.
Learning Period Per Scale for 12.5 Multiple UEBA Servers
Physical Machine (SERIES 6 ESA (DELL R640) SPECIFICATIONS)
Note: There are two UEBA servers, one is configured with Log and Endpoint data, while the other is configured with Network (TLS) data.
UEBA Server 1: 200,000 users with 150 million Log and Endpoint events per day
UEBA Server 2: 110 million Network events with 150,000 JA4 entities per day
|
Yes |
Fresh Installation of 12.5
Up to 12 days with 28 days of historical data
|
UEBA Server 1: 200,000 users with 150 million Log and Endpoint events per day
UEBA Server 2: 110 million Network events with 150,000 JA4 entities per day
|
Yes |
Upgrade from 12.2.x, 12.3, 12.3.1, 12.4, 12.4.1, and 12.4.2 to 12.5
UEBA reset is not required.
|
UEBA Server 1: 200,000 users with 150 million Log and Endpoint events per day
UEBA Server 2: 110 million Network events with 150,000 JA4 entities per day
|
Yes |
Note: An additional UEBA server must be installed to configure TLS schema on a separate UEBA server. For example, if you are planning to upgrade from 12.4: Before upgrade: UEBA Server 1 is configured with Log, Endpoint, and TLS data. After upgrade: • Add another UEBA Server 2 and configure it with TLS schema. • Reconfigure UEBA Server 1 with only Log and Endpoint data followed by UEBA reset. For more information, see Best Practices to Add and Remove Schemas for Multiple UEBA Servers.
Upgrade from 12.2.x, 12.3, 12.3.1, 12.4, 12.4.1, and 12.4.2 to 12.5 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required. Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
UEBA Server 1: 200,000 users with 150 million Log and Endpoint events per day
UEBA Server 2: 110 million Network events with 150,000 JA4 entities per day
|
No |
Fresh installation of 12.5
28 days to complete the learning period before generating alerts.
|
Learning Period Per Scale for 12.4
Physical Machine (SERIES 6 ESA (DELL R640) SPECIFICATIONS)
-
Scale 1: 125,000 users with 40 million Log and Endpoint events + 60 million Network events with 100,000 JA3 entities per day
-
Scale 2: 150,000 users with 50 million Log and Endpoint events + 70 million Network events with 150,000 JA3 entities per day
|
Yes |
Fresh Installation of 12.4
|
-
Scale 1: 125,000 users with 40 million Log and Endpoint events + 60 million Network events with 100,000 JA3 entities per day
-
Scale 2: 150,000 users with 50 million Log and Endpoint events + 70 million Network events with 150,000 JA3 entities per day
|
Yes |
Upgrade from 12.2.x, 12.3, 12.3.1 to 12.4
UEBA reset is not required.
|
-
Scale 1: 125,000 users with 40 million Log and Endpoint events + 60 million Network events with 100,000 JA3 entities per day
-
Scale 2: 150,000 users with 50 million Log and Endpoint events + 70 million Network events with 150,000 JA3 entities per day
|
Yes |
Upgrade from 12.2.x, 12.3, 12.3.1 to 12.4 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required.
Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
-
Scale 1: 125,000 users with 40 million Log and Endpoint events + 60 million Network events with 100,000 JA3 entities per day
-
Scale 2: 150,000 users with 50 million Log and Endpoint events + 70 million Network events with 150,000 JA3 entities per day
|
No |
Fresh installation of 12.4
28 days to complete the learning period before generating alerts.
|
Virtual Machine
The recommended vCPU specification for UEBA is Intel Xeon CPU @2.59 Ghz.
32 cores |
256GB |
192GB
|
-
Storage: 1.5 TB
-
Provisioning: Thick
|
500 |
500
|
To determine the scale limits for Virtual Machine deployments, refer to the Scaling Limitation Issue section in the Troubleshooting UEBA Configurations.
- You must reserve all the resources allocated to UEBA on the VM server. For example, if a user has a 2.1GHz CPU, then 32CPUs * 2.1GHz = 67.2GHz or 67200MHz must be reserved.
- The /var/netwitness partition must be mounted on a 1.5 TB Thick-provisioned disk for storage usage.
NetWitness recommends you to deploy UEBA on a virtual host, only if your log collection volume is low. If you have a moderate to high log collection volume, NetWitness recommends you to deploy UEBA on the physical host as described in the "NetWitness UEBA Host Hardware Specifications" topic of the Physical Host Installation Guide. Contact NetWitness Customer Support (https://community.netwitness.com/t5/support-information/how-to-contact-netwitness-support/ta-p/563897) for advice on choosing which host, virtual or physical, to use for UEBA.
-
Scale 1: 125,000 users with 40 million Log and Endpoint events + 20 million Network events with 100,000 JA3 entities per day
-
Scale 2: 150,000 users with 50 million Log and Endpoint events + 25 million Network events with 150,000 JA3 entities per day
|
Yes |
Fresh Installation of 12.4
|
-
Scale 1: 125,000 users with 40 million Log and Endpoint events + 20 million Network events with 100,000 JA3 entities per day
-
Scale 2: 150,000 users with 50 million Log and Endpoint events + 25 million Network events with 150,000 JA3 entities per day
|
Yes |
Upgrade from 12.2.x, 12.3, 12.3.1 to 12.4
UEBA reset is not required.
|
-
Scale 1: 125,000 users with 40 million Log and Endpoint events + 20 million Network events with 100,000 JA3 entities per day
-
Scale 2: 150,000 users with 50 million Log and Endpoint events + 25 million Network events with 150,000 JA3 entities per day
|
Yes |
Upgrade from 12.2.x, 12.3, 12.3.1 to 12.4 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required.
Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
-
Scale 1: 125,000 users with 40 million Log and Endpoint events + 20 million Network events with 100,000 JA3 entities per day
-
Scale 2: 150,000 users with 50 million Log and Endpoint events + 25 million Network events with 150,000 JA3 entities per day
|
No |
Fresh installation of 12.4
28 days to complete the learning period before generating alerts.
|
Network events per day refers to number of events consumed by UEBA per day.
Learning Period Per Scale for 12.4 Multiple UEBA Servers
Physical Machine (SERIES 6 ESA (DELL R640) SPECIFICATIONS)
There are two UEBA servers, one is configured with Log and Endpoint data, while the other is configured with Network (TLS) data.
Scale 1 for Multi-UEBA:
Scale 2 for Multi-UEBA:
- UEBA Server 1: 200,000 users with 150 million Log and Endpoint events per day
|
Yes |
Fresh Installation of 12.4
|
Scale 1 for Multi-UEBA:
Scale 2 for Multi-UEBA:
- UEBA Server 1: 200,000 users with 150 million Log and Endpoint events per day
|
Yes |
Upgrade from 12.2.x, 12.3, 12.3.1 to 12.4
UEBA reset is not required.
|
Scale 1 for Multi-UEBA:
Scale 2 for Multi-UEBA:
- UEBA Server 1: 200,000 users with 150 million Log and Endpoint events per day
|
Yes |
An additional UEBA server must be installed to configure TLS schema on a separate UEBA server. For example, if you are planning to upgrade from 12.2: Before upgrade: UEBA Server 1 is configured with Log, Endpoint, and TLS data. After upgrade: • Add another UEBA Server 2 and configure it with TLS schema. • Reconfigure UEBA Server 1 with only Log and Endpoint data followed by UEBA reset. For more information, see Best Practices to Add and Remove Schemas for Multiple UEBA Servers.
Upgrade from 12.2.x, 12.3, 12.3.1 to 12.4 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required. Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
Scale 1 for Multi-UEBA:
Scale 2 for Multi-UEBA:
- UEBA Server 1: 200,000 users with 150 million Log and Endpoint events per day
|
No |
Fresh installation of 12.4
28 days to complete the learning period before generating alerts.
|
Learning Period Per Scale for 12.3.1
Physical Machine (SERIES 6 ESA (DELL R640) SPECIFICATIONS)
125,000 users with 40 million Log and Endpoint events + 60 million Network events with 100,000 JA3 entities per day
|
Yes |
Fresh Installation of 12.3.1 Up to 12 days with 28 days of historical data
|
125,000 users with 40 million Log and Endpoint events + 60 million Network events with 100,000 JA3 entities per day |
Yes |
Upgrade from 11.7.x, 12.1.x, 12.2.x, 12.3 to 12.3.1
UEBA reset is not required.
|
125,000 users with 40 million Log and Endpoint events + 60 million Network events with 100,000 JA3 entities per day
|
Yes
|
Upgrade from 11.7.x, 12.1.x, 12.2.x, 12.3 to 12.3.1 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required.
Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
125,000 users with 40 million Log and Endpoint events + 60 million Network events with 100,000 JA3 entities per day
|
No
|
Fresh installation of 12.3.1
28 days to complete the learning period before generating alerts.
|
Virtual Machine
The recommended vCPU specification for UEBA is Intel Xeon CPU @2.59 Ghz.
32 cores |
256GB |
192GB
|
-
Storage: 1.5 TB
-
Provisioning: Thick
|
500 |
500
|
To determine the scale limits for Virtual Machine deployments, refer to the Scaling Limitation Issue section in the Troubleshooting UEBA Configurations.
- You must reserve all the resources allocated to UEBA on the VM server. For example, if a user has a 2.1GHz CPU, then 32CPUs * 2.1GHz = 67.2GHz or 67200MHz must be reserved.
- The /var/netwitness partition must be mounted on a 1.5 TB Thick-provisoned disk for storage usage.
NetWitness recommends you to deploy UEBA on a virtual host, only if your log collection volume is low. If you have a moderate to high log collection volume, NetWitness recommends you to deploy UEBA on the physical host as described in the "NetWitness UEBA Host Hardware Specifications" topic of the Physical Host Installation Guide. Contact NetWitness Customer Support (https://community.netwitness.com/t5/support-information/how-to-contact-netwitness-support/ta-p/563897) for advice on choosing which host, virtual or physical, to use for UEBA.
125,000 users with 40 million Log and Endpoint events + 20 million Network events with 100,000 JA3 entities per day
|
Yes |
Fresh Installation of 12.3.1 Up to 10 days with 28 days of historical data
|
125,000 users with 40 million Log and Endpoint events + 20 million Network events with 100,000 JA3 entities per day |
Yes |
Upgrade from 11.7.x, 12.1.x, 12.2.x, 12.3 to 12.3.1
UEBA reset is not required.
|
125,000 users with 40 million Log and Endpoint events + 20 million Network events with 100,000 JA3 entities per day |
Yes |
Upgrade from 11.7.x, 12.1.x, 12.2.x, 12.3 to 12.3.1 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required.
Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
125,000 users with 40 million Log and Endpoint events + 20 million Network events with 100,000 JA3 entities per day
|
No
|
Fresh installation of 12.3.1
28 days to complete the learning period before generating alerts.
|
Network events per day refers to number of events consumed by UEBA per day.
Learning Period Per Scale for 12.3.1 Multiple UEBA Servers
Physical Machine (SERIES 6 ESA (DELL R640) SPECIFICATIONS)
There are two UEBA servers, one is configured with Log and Endpoint data, while the other is configured with Network (TLS) data.
UEBA Server 1: 125,000 users with 120 million Log and Endpoint events per day
UEBA Server 2: 100 million Network events with 100,000 JA3 entities per day
|
Yes |
Fresh Installation of 12.3.1 Up to 12 days with 28 days of historical data
|
UEBA Server 1: 125,000 users with 120 million Log and Endpoint events per day
UEBA Server 2: 100 million Network events with 100,000 JA3 entities per day
|
Yes |
Upgrade from 11.7.x, 12.1.x, 12.2.x, 12.3 to 12.3.1
UEBA reset is not required.
|
UEBA Server 1: 125,000 users with 120 million Log and Endpoint events per day
UEBA Server 2: 100 million Network events with 100,000 JA3 entities per day
|
Yes |
An additional UEBA server must be installed to configure TLS schema on a separate UEBA server. For example, if you are planning to upgrade from 11.7.1: Before upgrade: UEBA Server 1 is configured with Log, Endpoint, and TLS data. After upgrade: • Add another UEBA Server 2 and configure it with TLS schema. • Reconfigure UEBA Server 1 with only Log and Endpoint data followed by UEBA reset. For more information, see Best Practices to Add and Remove Schemas for Multiple UEBA Servers.
Upgrade from 11.7.x, 12.1.x, 12.2.x, 12.3 to 12.3.1 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required. Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
UEBA Server 1: 125,000 users with 120 million Log and Endpoint events per day
UEBA Server 2: 100 million Network events with 100,000 JA3 entities per day
|
No
|
Fresh installation of 12.3.1
28 days to complete the learning period before generating alerts.
|
Learning Period Per Scale for 12.3
The displayed numbers are with the following enhancement enabled. Ensure that you enable the configuration in the application.properties file to improve the processing time. For more information, see The TLS model is taking too long to complete tasks section in the Troubleshooting UEBA Configurations.
Physical Machine (SERIES 6 ESA (DELL R640) SPECIFICATIONS)
100,000 users with 30 million Log and Endpoint events + 60 million Network events per day
|
Yes |
Fresh Installation of 12.3 Up to 10 days with 28 days of historical data
|
100,000 users with 30 million Log and Endpoint events + 60 million Network events per day |
Yes |
Upgrade from 11.7.x, 12.1.x, 12.2.x to 12.3
UEBA reset is not required.
|
100,000 users with 30 million Log and Endpoint events + 60 million Network events per day
|
Yes
|
Upgrade from 11.7.x, 12.1.x, 12.2.x to 12.3 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required.
Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
100,000 users with 30 million Log and Endpoint events + 60 million Network events per day
|
No
|
Fresh installation of 12.3
28 days to complete the learning period before generating alerts.
|
Virtual Machine
The recommended vCPU specification for UEBA is Intel Xeon CPU @2.59 Ghz.
32 cores |
128GB |
64GB
|
-
Storage: 1.5 TB
-
Provisioning: Thick
|
500 |
500
|
The /var/netwitness partition must be mounted on a 1.5 TB Thick-provisoned disk for storage usage.
To determine the scale limits for Virtual Machine deployments, refer to the Scaling Limitation Issue section in the Troubleshooting UEBA Configurations.
NetWitness recommends you to deploy UEBA on a virtual host, only if your log collection volume is low. If you have a moderate to high log collection volume, NetWitness recommends you to deploy UEBA on the physical host as described in the "NetWitness UEBA Host Hardware Specifications" topic of the Physical Host Installation Guide. Contact NetWitness Customer Support (https://community.netwitness.com/t5/support-information/how-to-contact-netwitness-support/ta-p/563897) for advice on choosing which host, virtual or physical, to use for UEBA.
100,000 users with 30 million Log and Endpoint events + 20 million Network events per day
|
Yes |
Fresh Installation of 12.3 Up to 8 days with 28 days of historical data
|
100,000 users with 30 million Log and Endpoint events + 20 million Network events per day |
Yes |
Upgrade from 11.7.x, 12.1.x, 12.2.x to 12.3
UEBA reset is not required.
|
100,000 users with 30 million Log and Endpoint events + 20 million Network events per day |
Yes |
Upgrade from 11.7.x, 12.1.x, 12.2.x to 12.3 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required.
Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
100,000 users with 30 million Log and Endpoint events + 20 million Network events per day
|
No
|
Fresh installation of 12.3
28 days to complete the learning period before generating alerts.
|
Network events per day refers to number of events consumed by UEBA per day.
Learning Period Per Scale for 12.3 Multiple UEBA Servers
Physical Machine (SERIES 6 ESA (DELL R640) SPECIFICATIONS)
There are two UEBA servers, one is configured with Log and Endpoint data, while the other is configured with Network (TLS) data.
UEBA Server 1: 100,000 users with 100 million Log and Endpoint events per day
UEBA Server 2: 100 million Network events per day
|
Yes |
Fresh Installation of 12.3 Up to 10 days with 28 days of historical data
|
UEBA Server 1: 100,000 users with 100 million Log and Endpoint events per day
UEBA Server 2: 100 million Network events per day
|
Yes
|
Upgrade from 11.7.x, 12.1.x, 12.2.x to 12.3
UEBA reset is not required.
|
UEBA Server 1: 100,000 users with 100 million Log and Endpoint events per day
UEBA Server 2: 100 million Network events per day
|
Yes
|
An additional UEBA server must be installed to configure TLS schema on a separate UEBA server. For example, if you are planning to upgrade from 11.7.1: Before upgrade: UEBA Server 1 is configured with Log, Endpoint, and TLS data. After upgrade: • Add another UEBA Server 2 and configure it with TLS schema. • Reconfigure UEBA Server 1 with only Log and Endpoint data followed by UEBA reset. For more information, see Best Practices to Add and Remove Schemas for Multiple UEBA Servers.
Upgrade from 11.7.x, 12.1.x, 12.2.x to 12.3 with schemas updated (Addition or removal of schemas configured on UEBA)
UEBA reset is required.
|
UEBA Server 1: 100,000 users with 100 million Log and Endpoint events per day
UEBA Server 2: 100 million Network events per day
|
Yes
|
Historical data is available for N days:
Historical data is not available
28 days to complete the learning period before generating alerts.
|
UEBA Server 1: 100,000 users with 100 million Log and Endpoint events per day
UEBA Server 2: 50 million Network events per day
|
No
|
Fresh installation of 12.3
28 days to complete the learning period before generating alerts.
|