Requirements for TrackMe
Target for deployments
For Splunk Enterprise customers
For Splunk on-premise customers, we recommend to deploy TrackMe on:
A dedicated Search Head or Search Head Cluster, especially if you intend to monitor a large number of environments or if you are a large scale Splunk customer
Alternatively, you can deploy TrackMe on the instance used for the purposes of the Splunk Monitoring Console
Splunk premium applications
We do not support colocating TrackMe with any of the Splunk premium applications (Splunk Enterprise Security, Splunk ITSI)
The reason is that both TrackMe and premium applications can have heavy requirements in term of workload, and TrackMe could therefore slightly impact these products
For Splunk Cloud customers
From Splunk Cloud Victoria, applications are deployed through self-services equally on all Search Heads, including Premium applications Search Head:
When deploying TrackMe on Splunk Cloud, the application will be deployed automatically on all Search Heads, including your Premium application Search Heads, if any
However, a virgin deployment of TrackMe has no workload as long as there are no Virtual Tenants created
TrackMe should be configured and used from the first Search Head tier of your Splunk Cloud stack, commonly called the Ad-hoc Search Head tier
Splunk premium applications
Similarly to on-premise deployments and while TrackMe is deployed on the premium applications Search Heads, we do not support running TrackMe on these Search Heads
The reason is that both TrackMe and premium applications can have heavy requirements in term of workload, and TrackMe could therefore slightly impact these products
System requirements for the use of TrackMe
Supported Operating Systems
TrackMe is supported on any flavour of:
Linux
MacOS
Currently, running TrackMe on Windows Operating Systems is not supported, although this might be the case in the future.
Containerized computing platforms
Similarly to Splunk Enterprise, running TrackMe on containerized environment is supported:
Resources requirement
The requirements for TrackMe in term of CPU slightly relies on its usage and configuration, the main factors are:
If the Splunk instance hosting TrackMe is itself more than a Search Head (if the instance is as well an indexer for instance in the case of all in one Splunk instance, which is not recommended)
The number of enabled TrackMe Virtual Tenants
The number of enabled Hybrid, Flex and Elastic Trackers
The number of entities monitored, which influence factors like the number of Machine Learning models to be created and maintained for instance
The repartitions between local tracking and Splunk Remote Searches tracking, which influence how much the load is shared between the local instance and the remote Splunk deployments, if any
The number of active users, the more users are actively using TrackMe, the less search slots are available for the back-end purposes
Minimum CPU requirements:
We recommend a minimal setup of 8 vCPUs for running TrackMe in a Production context
As a basis, a Virtual Tenant requires as an average 1 vCPU for running properly
Depending on the number of trackers, and their performance quality (for instance if use essentially high performing and well designed searches), a single Virtual Tenant can require more computes
A more recommended setup for running TrackMe at large scale in Production is 16 vCPUs ore more
Memory and Disk Performances:
TrackMe relies on Splunk and shares its typical requirements
We recommend to honour the minimal specifications for Splunk: https://docs.splunk.com/Documentation/Splunk/latest/Installation/Systemrequirements
Applications dependencies
TrackMe requires the following applications to be deployed and available on the instance hosting TrackMe:
Splunk_SA_CIM: https://splunkbase.splunk.com/app/1621
Splunk_ML_Toolkit: https://splunkbase.splunk.com/app/2890
semicircle_donut: https://splunkbase.splunk.com/app/4378
timeline_app: https://splunkbase.splunk.com/app/3120
note: The Splunk Machine Learning toolkit itself requires the Python Scientific package, TrackMe relies on the ML Toolkit for the management of Machine Learning models.
Shall any, or all of of the dependencies not being met, TrackMe will refuse to load the Virtual Tenant User Interface and show the list of missing applications, such as: