Splunk® User Behavior Analytics

Install and Upgrade Splunk User Behavior Analytics

This documentation does not apply to the most recent version of Splunk® User Behavior Analytics. For documentation on the most recent version, go to the latest release.

System requirements for Splunk UBA

Install Splunk UBA with assistance from Splunk Professional Services.

Hardware requirements

You can install Splunk UBA on a physical server, a virtual machine, or in the cloud.

Install Splunk UBA on its own hardware stack. Do not install Splunk UBA on the same machines as Splunk Enterprise.

Verify the following hardware requirements before installing Splunk UBA:

Disk space and memory requirements for installing Splunk UBA

Every machine in your Splunk UBA deployment must meet the following requirements. Not all machines in your deployment need to have matching specifications, but they must all meet the minimum requirements.

  • 16 CPU cores. If a machine has more than 16 cores, the additional cores are not used by Splunk UBA.
  • 64GB RAM. If a machine has more than 64GB RAM, the additional RAM is used by Splunk UBA as needed.
  • Disk 1 - 100GB dedicated disk space for the Splunk UBA installation on bare metal systems running RHEL, OEL, or CentOS. The OVA and AMI images are pre-configured with 50GB root disks for the Splunk UBA installation.
  • Disk 2 - 1TB additional disk space for metadata storage.
  • Disk 3 - 1TB additional disk space for each node running Spark services.

See Where services run in Splunk UBA in the Administer Splunk User Behavior Analytics manual for more information about how to determine where Splunk UBA services are running in your deployment. See Directories created or modified on the disk for more information about Splunk UBA directories and space requirements.

Do not manually mount the disks before installing Splunk UBA. During the installation procedure, the add-disk command will properly mount the disks for you.

The following table summarizes the disk requirements per deployment.

Splunk UBA Deployment Nodes Requiring 50GB Disk Space for Splunk UBA Nodes Requiring a 1TB Disk such as /var/vcap for Metadata Storage Nodes Requiring a 1TB Disk such as /var/vcap2 for Spark Services
1 Node Node 1 Node 1 Node 1
3 Nodes All Nodes All Nodes Nodes 1, 3
5 Nodes All Nodes All Nodes Nodes 1, 4, 5
7 Nodes All Nodes All Nodes Node 7
10 Nodes All Nodes All Nodes Nodes 9, 10
20 Nodes All Nodes All Nodes Nodes 17, 18, 19, 20

(Optional) Plan for configuring Splunk UBA warm standby

Configure warm standby in your deployment for high availability and disaster recovery. Allocate additional servers for a warm standby solution, where you can manually failover Splunk UBA to a full backup system. The backup system must have the same number of nodes as the active system. See Configure warm standby in Splunk UBA in Administer Splunk User Behavior Analytics.

(Optional) Add additional disks for offline Splunk UBA backups

Use the backup and restore scripts located in /opt/caspida/bin/utils to migrate your Splunk UBA deployment to the next larger size on the same operating system. For example, you can migrate from 5 nodes to 7 nodes, or 10 nodes to 20 nodes. If you want to migrate from 7 nodes to 20 nodes, migrate from 7 nodes to 10 nodes first, then from 10 nodes to 20 nodes.

Add an additional disk to the Splunk UBA management node mounted as /backup for the Splunk UBA backups used to restore Splunk UBA during the migration process.

The size of the additional disk must follow these guidelines:

  • The disk size must be at least half the size of your deployment in terabytes. For example, a 10-node system requires a 5TB disk.
  • If you are creating archives, allow for an additional 50 percent of the backup disk size. For example, a 10-node system requires a 5TB disk for backups, and an additional 2.5TB if for archives, so you would need a 7.5TB disk for archived backups.

The table summarizes the minimum disk size requirements for Splunk UBA backups per deployment:

Number of Splunk UBA Nodes Minimum Disk Size for Backup (without archives) Minimum Disk Size for Backup (with archives)
1 Node 1TB 1.5TB
3 Nodes 1TB 1.5TB
5 Nodes 2TB 3TB
7 Nodes 4TB 6TB
10 Nodes 5TB 7.5TB
20 Nodes 10TB 15TB

If you have previous backups on the same disk, be sure to also take this into account when determining available disk space. See Prepare to backup Splunk UBA in Administer Splunk User Behavior Analytics.

Supported AWS server instance types

If you run Splunk UBA on an AWS instance:

  • AWS measures CPU power on Elastic Compute Cloud (EC2) instances in virtual CPUs (vCPUs), not real CPUs.
  • Each vCPU is a hyper thread of an Intel Xeon core on most AWS instance types. See Amazon EC2 Instance Types on the AWS website.
  • As a hyper thread of a core, a vCPU acts as a core, but the physical core must schedule its workload among other workloads of other vCPUs that the physical core handles.

Installation of Splunk UBA on Amazon Web Services (AWS) servers is supported on the following instance types:

  • m4.4xlarge
  • m5.4xlarge
  • m5a.4xlarge
  • m5.8xlarge

All Splunk UBA nodes in your AWS environment must use io1 volumes for storage.

Disk subsystem IOPS requirements

For all new Splunk UBA deployments, the disk subsystem for each Splunk UBA server must support an average Input/Output Operations per second (IOPS) of 1200 IOPS. Existing deployments on 800 IOPS servers can be upgraded without having to upgrade the disks.

IOPS are a measurement of how much data throughput a hard drive can sustain. Because a hard drive reads and writes at different speeds, there are IOPS numbers for disk reads and writes. The average IOPS is the average between those two figures. See Disk subsystem in the Capacity Planning Manual for Splunk Enterprise for more about IOPS.

Network interface requirements

Splunk UBA requires at least one 1Gb ethernet interface on each node.

It is recommended that each Splunk UBA node is configured with at least one control plane interface and one data plane interface. Configure the control plane interfaces on one subnet, and the data plane interfaces on a separate subnet.

It is recommended that all interfaces on the data plane network be connected with at least one 10GbE or better ethernet interface. For larger clusters, use 25GbE, 40GbE or 50GbE network interfaces.

Install third-party agents after you install Splunk UBA

Third-party agents, such as anti-virus software, can block a successful installation of Splunk UBA. Install any desired third-party agents only after Splunk UBA is installed, and monitor Splunk UBA for any possible interference after the agents are installed and running.

If CPU utilization spikes when third-party agents are running on your UBA nodes, exclude the /var/vcap/store/docker/overlay2/ directory in the third-party agents setting.

Directories created or modified on the disk

Splunk UBA creates or modifies the following directories on the disk during installation.

Directory Disk Description of Contents Updated During Upgrade? Recommended Space
/home/caspida Disk 1 Contains the Splunk UBA installation and upgrade .tgz files. Yes 20 GB
/opt/caspida Disk 1 Contains the Splunk UBA software. Yes 10 GB
/opt/splunk Disk 1 Contains the Splunk forwarder to send data to the Splunk platform. Yes 10 GB
/etc/caspida/local/conf Disk 1 Contains custom configuration files affecting your local environment. No 1 GB
/var/vcap Disk 2 Contains the following notable sub-directories:
  • /var/vcap/packages - Contains Spark and Spark configuration files.
  • /var/vcap/sys/run - Contains files used by some Splunk UBA processes during initial startup.
  • /var/vcap/sys/log - Contains Splunk UBA log files.
  • /var/vcap/sys/tmp - Contains temporary files for some Splunk UBA processes.
  • /var/vcap/store - Contains the metadata storage for Splunk UBA services such as Kafka, PostgreSQL, Hadoop, Influxdb, and Redis
 Yes 1 TB
/var/vcap2 Disk 3 Contains the runtime data for Spark services. Yes 1 TB
/var Disk 1 Contains various support files required by Splunk UBA. The /var/lib directory must have a minimum of 20 GB. Yes 50 GB

Operating system requirements

You must install Splunk UBA on a server that uses one of the following operating systems:

Installing Splunk UBA on hardened operating systems is not supported.

Operating System Kernel Version Tested Installation Type and Description
Red Hat Enterprise Linux (RHEL) 7.9 Basic Server (Maipo) Linux-3.10.0-1160.36.2.el7.x86_64 Bare metal on supported hardware. Obtain the software from Splunk UBA Software Installation Package on Splunkbase.
CentOS 7.9.2009 3.10.0-1160.36.2.el7.x86_64
Red Hat Enterprise Linux (RHEL) 7.8 Basic Server Linux-3.10.0-1127.el7.x86_64-x86_64-with-redhat-7.8-Maipo
CentOS 7.8 Linux-3.10.0-1127.el7.x86_64-x86_64-with-centos-7.8.2003-Core
Oracle Enterprise Linux (OEL) 7.7 Linux-4.14.35-1902.300.11.el7uek.x86_64-x86_64-with-oracle-7.7
Red Hat Enterprise Linux (RHEL) 7.7 Basic Server Linux-3.10.0-1062.12.1.el7.x86_64-x86_64-with-redhat-7.7-Maipo
CentOS 7.7 Linux-3.10.0-1062.4.3.el7.x86_64-x86_64-with-centos-7.7.1908-Core
Ubuntu 16.04.3 LTS Linux-4.4.0-176-generic-x86_64-with-Ubuntu-16.04-xenial This operating system is bundled with the OVA and AMI installation packages.
  • The OVA package is available for VMware environments. You can obtain the OVA package from Splunk UBA OVA Software on Splunkbase.
  • The AMI package is available for AWS environments and can be shared with your AWS account upon request.

Splunk UBA requires that the operating system and underlying component versions match exactly on all nodes in your deployment. Updating the operating system or any components in your deployment can break dependencies that will cause Splunk UBA to stop working and is not recommended. If you must update the operating system before the next release of Splunk UBA, do so in a test environment and verify that everything is working properly before applying the upgrade to your production environment.

Additional RHEL requirements

Make sure your RHEL server has access to the RHEL repositories, and the license includes the following subscription names:

  • Red Hat Enterprise Linux Server
  • Red Hat Enterprise Linux Server - Extended Update Support (EUS)

The RHEL EUS subscription enables you to remain with previous stable releases of RHEL for up to approximately 24 months.

Applying security patches to your operating system

Always apply the latest security patches for your operating system on all servers in your deployment.

Use the following command to apply all available security patches: 

yum update --security -y

Use the following command to upgrade only those packages with security errata: 

yum update-minimal --security -y

Do not apply any non-security related patches such as bug fixes or version updates to your operating system to any Splunk UBA servers.

User access requirements

If you are installing Splunk UBA using an OVA or AMI image, perform all tasks as the caspida user and use sudo for tasks requiring root-level privileges.

If you are installing Splunk UBA on a supported Linux platform, you must be able to do the following:

  • Be able to log in as root, or log in as a different user and use su or sudo to have root privileges. This is required for preparing the servers prior to installing the Splunk UBA software.
  • Create the caspida user with the appropriate privileges. The caspida user is required to install the Splunk UBA software.
  • All user and group authentication must be performed locally on each Splunk UBA host. Authenticating users and groups using a centralized controller or user and group management system is not supported.

Networking requirements

Perform the following tasks or verify specific information to meet the networking requirements for installing Splunk UBA:

Assign static IP addresses to Splunk UBA servers

Assign static IP addresses to Splunk UBA servers.

Inbound networking port requirements

Splunk UBA requires the following ports to be open to the outside world for other services to interact with Splunk UBA.

Service Port
SSH 22
HTTPS 443

Splunk UBA requires the following ports to be open internally among the nodes in a distributed Splunk UBA cluster to allow specific services to interact with each other.

Splunk UBA requires port 9093 to be an open inbound networking port from Splunk Indexers in a distributed Splunk UBA cluster if Kafka data ingestion is used.

Service Port
SSH 22
Redis 6379, 16379
PostgreSQL 5432
Zookeeper 2181, 2888, 3888
Apache Kafka 9092, 9901, 9093 (for Kafka ingestion), 32768 - 65535 (for JMX)
Job Manager 9002
Time Series Database 8086
Apache Impala 21000, 21050, 25000, 25010, 25020
Apache Spark 7077, 8080, 8081
Hadoop Namenode 8020
Hadoop Namenode WebUI 50070
Hadoop Yarn ResourceManager 8090
Hadoop Data Transfer Port 50010
Hadoop Datanodes 50020, 50075
Hadoop Secondary namenode 50090
Hive Metastore 9090, 9095
Kubernetes/etcd 2379, 2380, 5000, 6443, 10250, 10251, 10252, 10255, 30000 - 32767

For more details on services in Splunk UBA, see Monitoring the health of your Splunk UBA deployment in Administer Splunk User Behavior Analytics.

Splunk platform port requirements

The following ports must be open on the Splunk platform to interact with Splunk UBA:

Service Port
HTTPS authentication 443
HTTP authentication 80
REST services that enable Splunk UBA to communicate with Splunk search heads 8089
Port used to send alerts to Splunk Enterprise Security (ES) User-defined (for example, 10008)

Modify firewalls and proxies

Modify firewalls and proxies to support the inbound and outbound port requirements defined in this document so that requests to internal services do not attempt to travel externally.

  • Set the no_proxy environment variable for general HTTP communication between nodes.
  • Set the NO_PROXY environment variable for Splunk UBA's time series database (influxdb). Set NO_PROXY to the same values as no_proxy.

Perform the following tasks to configure your firewall and proxy settings:

  1. If you use an HTTP or HTTPS proxy, exclude localhost and the IP addresses and names of the Splunk UBA servers from the proxy. For example, in a 3-node cluster, add the following configuration to the /etc/environment file: 
    # Proxy host/port for reference. These variables are not used below. 
PROXY_IP="1.2.3.4"
PROXY_PORT="3128"

# Set the proxy variables based on the values above. Both upper and lower case, different services look for different casing.
HTTP_PROXY="http://1.2.3.4:3128"
http_proxy="http://1.2.3.4:3128"

HTTPS_PROXY="https://1.2.3.4:3128"
https_proxy="https://1.2.3.4:3128"

# Exclude loopback addresses from the proxy
# Note: CIDR ranges aren't supported by older tools so specify both IP and CIDR
#
# Proxy values to be set:
# localhost: "localhost,127.0.0.1,127.0.1.1,127.0.0.0/8"
# UBA Containers: 10.96.0.0/12,10.244.0.0/16,172.17.0.1,172.17.0.2,172.17.0.0/16"
# Site Specific hosts by shortname, fqdn, ip: "ubanode01,ubanode01.mydomain.local,10.10.10.1", ubanode02,ubanode02.mydomain.local,10.10.10.2""

# Set NO_PROXY and no_proxy
NO_PROXY="localhost,127.0.0.1,127.0.1.1,127.0.0.0/8, 10.96.0.0/12,10.244.0.0/16,172.17.0.1,172.17.0.2,172.17.0.0/16, ubanode01,ubanode01.mydomain.local,10.10.10.1, ubanode02,ubanode02.mydomain.local,10.10.10.2, ubanode03,ubanode03.mydomain.local,10.10.10.3"

no_proxy="localhost,127.0.0.1,127.0.1.1,127.0.0.0/8, 10.96.0.0/12,10.244.0.0/16,172.17.0.1,172.17.0.2,172.17.0.0/16, ubanode01,ubanode01.mydomain.local,10.10.10.1, ubanode02,ubanode02.mydomain.local,10.10.10.2, ubanode03,ubanode03.mydomain.local,10.10.10.3"
  2. Use the following commands to stop and restart all Splunk UBA services for the changes in /etc/environment to take effect: 
    /opt/caspida/bin/Caspida stop-all
/opt/caspida/bin/Caspida start-all
  3. Verify that the nslookup localhost command returns a 127.x.x.x IP address. For example: 
    $ nslookup localhost
Server:		10.160.20.4
Address:	10.160.20.4#53

Name:	localhost.sv.splunk.com
Address: 127.0.0.1

Configure host name lookups and DNS

Configure your environment so that Splunk UBA can resolve host names properly.

Configure the name switching service

The name switching service in Linux environments determines the order in which services are queried for host name lookups. Use cat /etc/nsswitch.conf to verify that your name switching service is using files before DNS. Check the hosts line in the output:

  • If you see files dns it means that /etc/hosts will be queried before checking DNS.
  • If you see dns files it means that DNS will be queried before the /etc/hosts file.

Also make sure myhostname is the last item on the hosts line so that the system can determine its own host name from the local config files. 

$ cat /etc/nsswitch.conf
# /etc/nsswitch.conf
#
# Example configuration of GNU Name Service Switch functionality.
# If you have the `glibc-doc-reference' and `info' packages installed, try:
# `info libc "Name Service Switch"' for information about this file.

passwd:         compat
group:          compat
shadow:         compat
gshadow:        files

hosts:          files dns myhostname
...

Configure the DNS resolver

Some Splunk UBA services use DNS during installation and while the product is running. All nodes in your Splunk UBA deployment must point to the same DNS server. Verify this is the case in the /etc/resolv.conf file on each node. Use the following command to check if /etc/resolv.conf exists on your system: 

ls -lH /etc/resolv.conf

If the file does not exist, create the file by performing the following tasks:

  1. Run the following command: 
    sudo systemctl enable resolvconf
  2. Restart the server.
  3. Run the ls -lH /etc/resolv.conf command again to verify that the /etc/resolv.conf exists.

Verify the network interface configuration

Verify that the network interface configuration has a dns-search value configured to match your domain, such as mgmt.corp.local. Check the /etc/resolv.conf file to see if search mgmt.corp.local is present so that any shortname lookups for other local nodes are resolved correctly.

  • On Ubuntu systems, the configuration is located in /etc/network/interfaces as: 
    dns-search mgmt.corp.local
  • On CentOS, RHEL, and Oracle Linux systems, the configuration may be located in /etc/sysconfig/network-scripts/ifcfg-eth0 as: 
    DOMAIN=mgmt.corp.local
    More recent CentOS, RHEL, and Oracle Linux systems may use a different slot-based naming scheme. The exact name may vary depending on your specific environment.

Be consistent with your naming conventions and use either all fully qualified domain names (FQDN) such has host.example.com or all short names such as host. Do not use FQDNs in some places and short names in others.

Configure local DNS using the /etc/hosts file

Verify that the /etc/hosts file identifies each node in your Splunk UBA cluster using the following format: 

<IP address> <FQDN> <short name> <alias>

For example: 

192.168.10.1    spluba01.mgmt.corp.local    spluba01    ubanode01
192.168.10.2    spluba02.mgmt.corp.local    spluba02    ubanode02
192.168.10.3    spluba03.mgmt.corp.local    spluba03    ubanode03
192.168.10.4    spluba04.mgmt.corp.local    spluba04    ubanode04
192.168.10.5    spluba05.mgmt.corp.local    spluba05    ubanode05

In this example, host spluba01 has an IP address of 192.168.10.1 and its FQDN is spluba01.mgmt.corp.local. Anything after the first three field is considered an alias, and is optional. In this example, we use ubanode1 is used to identify node number 1, ubanode2 is used to identify node number 2, and so on.

Formatting your /etc/hosts file this way in conjunction with using files before DNS in /etc/nsswitch.conf means that both short names and FQDNs can be obtained without any DNS lookups.

If you choose to not include the FQDN in the /etc/hosts file, you must add the domain name into the /etc/resolv.conf file in order for DNS to work properly in your environment.

Verifying your name lookup and DNS settings

Test your name lookup and DNS settings to make sure you get the expected output.

  • Use various hostname commands and verify the expected output. For example, from the spluba01.mgmt.corp.local node: 
    $ hostname
spluba01
$ hostname -s
spluba01
$ hostname --fqdn
spluba01.mgmt.corp.local
  • Use the ping <short name> command from each Splunk UBA node to all other Splunk UBA nodes and verify that all nodes can be reached.
  • Use the ping <FQDN> command from each Splunk UBA node to all other Splunk UBA nodes and verify that all nodes can be reached.

Supported web browsers

Open Splunk UBA in the latest versions of any of the following web browsers. Splunk UBA does not support other web browsers, such as Internet Explorer.

  • Mozilla Firefox
  • Google Chrome
  • Apple Safari

Supported single sign-on identity providers

Splunk UBA supports single sign-on integration with the following identity providers:

  • Ping Identity
  • Okta
  • Microsoft ADFS
  • OneLogin

See Configure authentication using single sign-on in Administer Splunk User Behavior Analytics.

Requirements for connecting to and getting data from the Splunk platform

To send data from Splunk platform to Splunk UBA, you must have specific Splunk platform versions and a properly configured user account. See Splunk UBA product compatibility matrix in the Plan and scale your Splunk UBA Deployment manual.

Requirements for the Splunk Enterprise user account

If you create a custom role, and the user in that role handles information related to data sources, make sure that custom role has permissions to edit_roles or admin_all_objects. These permissions allow the custom role user to drill down into event queries.

Verify that you have a Splunk Enterprise user account with:

  • Capabilities to perform real-time search, perform REST API calls, and access to the data. The admin role in Splunk Enterprise has the required capabilities by default. If you use a different role, you need the rt_search, edit_forwarders, list_forwarders, and edit_uba_settings capabilities. Add these capabilities to a role in Splunk Web. See Add and edit roles with Splunk Web in Securing Splunk Enterprise.
  • Configure the search job limits for the Splunk Enterprise user account and role so that they are twice the number of maximum allowed data sources for your deployment.
    Size of cluster Max number of data sources User-level concurrent search job limit User-level concurrent real-time search job limit Role-level concurrent search job limit Role-level concurrent real-time search job limit
    1 node 6 12 12 12 12
    3 nodes 10 20 20 20 20
    5 nodes 12 24 24 24 24
    7 nodes 24 48 48 48 48
    10 nodes 32 64 64 64 64
    20 nodes 64 128 128 128 128
  • Configure the Splunk Enterprise user account to have sufficient disk usage quota (for example, 40GB).

Send data to and receive data from Splunk Enterprise Security

To send and receive data from Splunk Enterprise Security, you must have the Splunk add-on for Splunk UBA installed and enabled on your search head with the ueba index deployed to your indexers. See Splunk UBA product compatibility matrix in the Plan and Scale your Splunk UBA Deployment manual for information about version compatibility among products.

Splunk Cloud Platform customers must contact Splunk Support to fully integrate with Splunk UBA. The Splunk Cloud Platform admin role cannot perform Splunk UBA setup.

Send data from Splunk Enterprise directly to Kafka in Splunk UBA

Use the Splunk UBA Kafka Ingestion App to send data from large data sets in Splunk Enterprise directly to Kafka in Splunk UBA. Sending data directly to Kafka offloads the processing task from the search heads to the indexers. See Requirements for Kafka data ingestion in the Splunk UBA Kafka Ingestion App manual.

Monitor Splunk UBA directly from Splunk Enterprise

Use the Splunk UBA Monitoring App to monitor the health of Splunk UBA and investigate Splunk UBA issues directly from Splunk Enterprise. See Splunk UBA Monitoring app requirements in the Splunk UBA Monitoring App manual.

Installing Splunk UBA in environments with no Internet access

Some environments require Splunk UBA to be installed without access to the Internet. In such cases, the functionality of Splunk UBA will be limited in the following areas:

  • Splunk UBA pages that normally show visual geographical location information about a device will show warnings that the Google Maps API cannot be reached. Perform the following tasks to disable Splunk UBA from using geographical location and displaying the warning:
    • In Splunk UBA, select Manage > Settings.
    • Select Geo Location.
    • Deselect the checkbox in the Show Geo Maps field.
    • Click OK.
  • Clicking the Learn more link on any Splunk UBA page will open a new tab with a link to quickdraw.splunk.com. This is the URL used to generate the correct help link to the Splunk UBA documentation.
Last modified on 15 December, 2023
Splunk UBA installation checklist   Check system status before and after installation

This documentation applies to the following versions of Splunk® User Behavior Analytics: 5.0.5, 5.0.5.1

Acrobat logo Download manual
Acrobat logo Download this page

Was this topic useful?







You must be logged into splunk.com in order to post comments. Log in now.

Please try to keep this discussion focused on the content covered in this documentation topic. If you have a more general question about Splunk functionality or are experiencing a difficulty with Splunk, consider posting a question to Splunkbase Answers.

0 out of 1000 Characters