Pre-requisites (On-premise installation)
oepn
This document has been created to simplify communication between your IT team and 3YOURMIND. It will also help you to better understand the requirements for your IT infrastructure.
We support installing our application with 2 methods:
If you have an existing Kubernetes Cluster and have experience with Kubernetes or Red Hat Open Shift Cluster, see:
Installation into existing Kubernetes ClusterIf you don’t have an existing cluster, see:
Installation on Bare Metal Server
Installation into existing Kubernetes Cluster
If possible, we recommend installing our application into an existing Kubernetes or Red Hat Open Shift cluster.
Installing and operating a Kubernetes Cluster is often underestimated. We recommend this option only if a multi-purpose cluster already exists and is not built just for 3YOURMIND.
Cluster Requirements
Minimum Resources (to run the SaaS M resources equivalent):
4 CPU
Intel Broadwell (2013) or newer
AMD Piledriver (2012) or newer
16 GB RAM
Recommended Resources (To run the SaaS L equivalent):
8 CPU
32 GB RAM
IngressController (should support SSL/TLS)
SSL/TLS Certificates
StorageClass (It must be possible to create Persistent Volumes)
Space for Uploads and Media Files on volume(see Disk Space) Either:
On Persistent Volumes
On AWS S3
External database: PostgreSQL 13 or later
DNS
SMTP Server
Installation on Bare Metal Server
Our application can be installed on any bare-metal server.
Single K8S cluster software is included with the installation package for bare metal servers. No need to provision any of the Kubernetes components.
Bare-Metal server Requirements
Minimum Resources (to run the SaaS M resources equivalent):
4 CPU
Intel Nehalem (2008) or newer
AMD Bulldozer (2011) or newer
16 GB RAM
Recommended Resources (To run the SaaS L equivalent):
8 CPU
32 GB RAM
Space for Uploads and Media Files on volume(see Disk Space) Either:
On Persistent Volumes/local disk storage
On AWS S3
Minimum 120 GB of disk space for the
/rootpartition.min 60 GB in
/var/lib/min 10 GB in
/var/log/min 10 GB in
/usr/libexec/min 20 GB in
/opt/
External database: PostgreSQL 11 or later
DNS
SMTP Server
Supported operating systems
The software is tested on Ubuntu 22.04
The software is also known to run on RHEL 8 and likely runs on other linux-based Operating Systems as well.
Software Requirements
Bash Version 4 or greater
Root access (no shared instance)
Network Requirements
The server must be connected to a network (e.g. a router). The software does not work on a stand-alone device without network interface.
To access the application from other clients, it is further required:
The server must have valid DNS record and hostname.
All hosts in the cluster must have static IP address assignments on a network interface that will be used for routing to containers
SSL/TLS Certificates
Disk Space
Since the application deals with 3D Data and CAD files, a lot of data can be produced.
The media files are stored at a directory, which can be specified during the installation.
The directory could also be a separate hard drive or network share which is mounted on the linux file system.
The needed space is highly dependent on the usage of the application and the average size of the uploaded 3D Data. We recommend clarifying the needs early and reserving enough Space on a Hard Drive.
Also we support storing uploaded files on AWS S3.
Firewall requirements
The following Ports must be accessible:
Web: TCP Port 80/443 (for all users)
SSH: TCP Port 22 (for maintenance)
Additionally, the Linux Server must be capable to connect to
the PostgreSQL Database server
the SMTP Server
An internet connection is not required.
Database Requirements
The Application is tested with a PostgreSQL 17 (Encoding UTF-8). It will likely also work with PostgreSQL 13 or newer versions.
Details:
The Application requires a separate PostgreSQL database server.
The application server needs read/write access to the database server, so any firewall between these needs to allow traffic (Default Port 5432)
3YOURMIND will and can not manage the database server for you. We will neither install a database server nor configure updates/replication of existing database servers.
Sizing requirements for the database heavily depend on usage. We recommend at least 20 GB for a production system. The database will grow slowly over time, so keep the disk space under monitoring.
If the database is installed manually:
It should be done on a separate server with at least 4GB RAM
The Application Server must be able to connect to the Database Server (Default Port 5432)
For installing PostgreSQL, please refer to the official Installation instructions or to an unofficial installation guide (recommended)
After Installation
On your database server, you will need to create one database and one user, which has read/write/drop/create rights on the said database.
Set these variables for the application user on the database (exchange
<dbuser>with your database username):ALTER ROLE <dbuser> SET client_encoding TO 'utf8'; ALTER ROLE <dbuser> SET default_transaction_isolation TO 'read committed'; ALTER ROLE <dbuser> SET timezone TO 'UTC';
AI capabilities requirements (optional)
In order to use our AI capabilities to either extract information from 2D technical drawings, use our material mapper and enhanced part evaluation/retrieval you will need a connection to a vision capable LLM model.
Information like uploaded 2D technical drawings, part metadata (such as material name) will be transferred to the LLM provider
Bring Your Own Token (BYOT) - Recommended
Requires a network connection to an internal LLM provider or an internet connection to a cloud hosted one.
We support all the major LLM providers (see this list) as well as existing LLM deployments within your organization. We recommend openai gpt-4.1 or better.
Make sure the following are available:
A LLM provider token if available:
Example: My organization has a Claude cloud account → token must be provided
Example: My organization has an on-Prem deployed model for internal use → token may not be needed
The LLM provider base URL if model is not reached through its default URL:
Example: My organization uses OpenAI models on OpenAI Cloud →
no base URL needs to be spacifiedExample: My organization uses OpenAI models on Azure Cloud → base URL must be provided
The provider/model name such as
openai/gpt-4.1
Embedded Deployment
Requires a gpu enabled server.
We also support embedded deployments of LLMs through ollama. The supported models can be found here. Recommended models and server requirements can be found on the table below.
Server requirements:
| Minimum (w/o technical drawings) | Recomended (with Technical Drawings) |
vRAM | 16GB | 80GB |
example GPU | Nvidia A10,Nvidia A2000, Nvidia Quadro rtx 5000, Nvidia Quadro rtx pro 2000 | Nvidia Tesla A100, Nvidia Quadro rtx pro 6000 blackwell, 2x Nvidia L40S |
Server estimated cost (one time) | $5,000 to over $10,000 | $12,000 to $25,000+ |
Server estimated on-demand cost (e.g. AWS) | $300 to $500 per month | $2700 to $4200 per month |
Recommended AI Models | gemma3:12b | qwen2.5:72b |
Prerequisites:
nvidia-driver-580(or any driver compatible with your graphics accelerator)nvidia-container-toolkit
Machine Connectivity Requirements (optional)
Please see Machine Connectivity System Requirements.
Other Requirements
To maintain a stable 3YOURMIND Application on the server, the following tasks need to be performed:
Provisioning all above resources
Setting up a Database System and executing Queries (SQL)
Configuring DNS
Setting up SSL
Installing and (regularly) updating the 3YOURMIND Software
In order to perform these tasks, the executing person needs the correct access rights and some knowledge of SSH, Bash on Linux (ls, cd, mv, cp, chown, chmod, nano/vi, top, etc.) Docker and or Kubernetes knowledge is a benefit.
The person should be responsible for both installing the software AND operating the software productively mode (monitoring, error inspection, incident handling, upgrading, …).