Skip to main content


Modifying Server Memory Settings

When adding memory to or removing memory from a Brainspace server, you must modify the application configurations to recognize and to enable server memory for Brainspace, and memory settings must be updated to take advantage of the modified resources available on the affected Brainspace servers. This applies to all Brainspace servers.

The relevant settings should be modified through the environment variables set for the user brains in ~/.bash_profile by updating the settings based on the instructions in this article and then restarting the services. For the On-Demand and Analytics servers, restart the brainspace-analysis service while on the Application server. The Tomcat service must also be restarted.


This configuration change should only be made by advanced users. Please consult Brainspace Support before making any environment variables changes.





Applies to all Brainspace Discovery servers

This is the amount of memory allocated to active datasets and to run Batch Tools commands.

BT_HEAP_MAX should be set to 70% of total server RAM.

Ensure the OS has sufficient RAM and does not need to use SWAP. The need for additional OS RAM can be aggravated by additional SW packages beyond 'Basic server'.


Application (Runtime) server only

This is the amount of memory allocated to the tomcat web service for the Discovery user interface (UI).

CATALINA_OPTS should be set to 70% of the total server RAM.

The Lucene index for each dataset is loaded into RAM outside of CATALINA_OPTS. This means that a large, complex or 'rich' dataset can end up using additional RAM. Additional RAM can be added or cannibalized from CATALINA_OPTS, without needing to change the environment settings.


This is the workspace size for the java process in the analytics that support focuses and initial analysis. This includes the graph index, email and domain index, keymap, cluster and content data, the brain and other supporting information.

ANALYSIS_HEAP_MAX should be set to 4096M, with BT_HEAP_MAX getting most of the remaining RAM. Be sure to leave sufficient RAM unallocated for the OS).

Note: Very large focuses, from large data populations, may require more RAM allocated to ANALYSIS_HEAP_MAX. The focus process also uses BT_HEAP_MAX. So, if the ANALYSIS_HEAP_MAX setting gets disproportionately large you may need to drop the associated BT_HEAP_MAX.

Memory allocation (ratio) between Java & Discovery (recommended: 70%).

Please consult Brainspace Support before changing.