Skip to main content

Brainspace

Latest Release Announcement

Reveal-Brainspace is improving underlying IT architecture to make it easier to manage operating system (OS) version changes going forward. Starting with the Brainspace 7.0 release, we are introducing a Linux OS that supports Docker, and more specifically, we recommend Ubuntu LTS as the underlying OS supporting Brainspace.

This guide is designed to help our clients quickly and easily deploy or migrate the Brainspace application to the 7.0 Release planned for the first quarter of 2024. Brainspace currently supports Chrome, Edge (Chromium-based versions) or Firefox browsers installed with WebGL enabled.

image1.png
image2.png

If you have any questions, please contact Brainspace Support at for assistance.

Brainspace 7.0 Server Guidelines

Server Resource recommendations vary depending on client data, document and dataset size, complexity and the Brainspace features being used.

Use these server recommendations as a guideline only: monitor system resource utilization closely and update as required.

If using VM instances, use Resource Allocation Reservations to ensure committed server performance.

image3.png

For larger volume instances, please consult your Brainspace Solutions Architect or submit a request to .

Additional Sizing Considerations

  • The Brainspace Server Guidelines table above will provide full Brainspace functionality and performance for a wide range of document populations. Actual resource needs will vary depending on source documents complexity and the combination of Brainspace analytics being used.

  • Brainspace processes all extracted text and metadata without a requirement for pre-filtering, however large, complex data types can reduce system performance and require additional resources.

  • Larger datasets tend to consume more RAM than multiple smaller datasets due to larger brains and more complex relationships within the document population.

  • Server specific resource considerations.

    • Application Server – All Active datasets are loaded into RAM for optimal graphical response. Large documents and datasets will increase memory requirements. Memory and Processor utilization is driven by Concept Search, Cluster Wheel and Communications Analysis activity.

    • Analytics Server – Only one dataset can be built at a time on the Analytics server. Additional Analytics servers can be deployed to allow for multiple, parallel dataset builds to occur. Additional resources (RAM & Processors, always at least a 4:1 ratio) may be required for very large dataset builds or may help improve dataset build times.

    • On-Demand Analytics (ODA) Server – Only one Focus can be built at a time on the ODA server. Additional ODA servers can be deployed to allow for multiple, concurrent Focus builds. Also, machine learning activity for CMML and PC occur on the ODA server and heavy classifier users should increase system resources accordingly. Additional resources (RAM & Processors, always at least a 4:1 ratio) may be required for very large Focus builds or may help improve CMML, PC and Focus build times.

  • Increased user activity in Brainspace does not drive system resources quite as much as data and system activity. Plan accordingly for dataset volume (number and size of documents) and active projects concurrently in the system. A good rule of thumb for high user counts is to add 0.5GB per active user (i.e. 20GB for 40 concurrently logged in users) to the Application Server.

  • Disk usage on the /data volume will expand by up to six times the size of your source text during the Brainspace dataset build process, index creation and optimization. Following the dataset completion and cleanup process the data will typically occupy from 1-2x the source text on disk.

  • Keep in mind that usage requirements will likely change between minor versions (7.1 to 7.2, 7.2 to 7.3, etc.) as new features are added.

Disk Layout

Brainspace recommends 100GB of disk storage for each server, utilizing the following layout. Please note that any deviations from this recommendation may impact upgrades or application performance and should be discussed with the Brainspace Support PM or Engineer prior to upgrade or install. Any file storage for client-managed logging or security applications should be in addition to this recommendation.

The /var partition volume size should be, at minimum, 100GB, and ideally equal to the /localdata size on the Application server in very large environments with tens of millions of documents. The On-Demand Analytics server does its work in a local Docker volume stored in /var/lib/docker.

The /data volume should be a high-performance storage solution. It can be a disk on the Application server that gets shared, using NFS or other supported protocol, to the other servers, or it may be an NFS share from an Enterprise storage solution with minimum 10,000 IOPS and at least 1Gbps actual throughput between servers.

The /localdata volume should be a high-performance storage solution. It can be a disk on the Application server that gets shared, using NFS or other supported protocol, to the other servers, or it may be an NFS share from an Enterprise storage solution with minimum 10,000 IOPS and at least 1Gbps actual throughput between servers.

The NFS share must be configured before install, just like /data.

You will need the IP address of the NFS server and the NFS protocol version handy while performing the install.

This is our recommended disk layout. Alternatives can be configured if desired, so long as the general layout remains as follows:

Mount Point

Volume Size

Application Server

/boot

250 M

/var

100G minimum

/tmp

20G

/

Remainder

/localdata (NFS that will be shared)

Variable

/data (NFS that will be shared)

Variable

Analytics Server / On Demand Server

/boot

250 M

/var

100G minimum

/tmp

20G

/

Remainder

/localdata

Variable