Infrastructure IA

Ressources de calcul

Le cluster du laboratoire combine virtualisation, infrastructure conteneurisée, serveurs GPU pour l’inférence et nœuds pour le calcul parallèle et distribué.

Avantages étudiants Resource monitoring
Salle serveur du laboratoire IA de la LNU
Server-room infrastructure of the LNU AI laboratory: virtualization, GPU inference nodes and cluster services for education and research.
01

Proxmox cluster

The virtualization layer for educational and research services, isolated environments, GPU nodes and rapid virtual machine deployment.

25 servers with virtual machines
02

Rocks Cluster

A dedicated cluster for parallel and distributed computing, simulations, batch data processing and educational HPC practice.

8 nodes for HPC workloads
03

Kubernetes

Container orchestration for laboratory services, experimental APIs, educational environments and scalable applied systems.

service and API orchestration
04

Docker ecosystem

A consistent approach to reproducible environments: containers for Python, ML stacks, databases, queues, web services and student projects.

reproducible environments
05

AI inference servers

GPU servers for computer vision, NLP, generative AI and applied inference scenarios in education and research.

10 standalone inference servers
Infrastructure map

Cluster topology

A compact operational view of virtualization, GPU inference, HPC nodes and container services in the laboratory infrastructure.

Infrastructure IA 43 nœuds
Proxmox 25 serveurs
Inférence IA 10 serveurs autonomes
Rocks 8 nœuds HPC
Kubernetes service orchestration
Docker ecosystem AI inference servers University AI infrastructure
Cluster resources

Resource monitoring

25 Proxmox servers

virtualization, student VMs, GPU nodes

8 Rocks nodes

parallel and distributed computing

10 Inference servers

standalone servers for AI model serving

198 CPU cores

rounded total compute capacity

490 GB RAM

approximate total memory capacity

1% CPU

198 cores

total CPU capacity
41% Memory

201 GB / 490 GB

used
16% Storage

895.77 GiB / 5.34 TiB

used

Proxmox

25 en ligne
pxm-01 en ligne

GPU inference

CPU + GPU
18 VM 42%
RTX 3080 Ti
pxm-02 en ligne

GPU training

CPU + GPU
14 VM 57%
2x RTX 3080
pxm-03 en ligne

Virtualization

CPU
12 VM 34%
pxm-04 en ligne

Legacy GPU

CPU + GPU
9 VM 36%
GTX 1080 Ti
pxm-05 en ligne

Virtualization

CPU
16 VM 31%
pxm-06 en ligne

Virtualization

CPU
11 VM 28%
pxm-07 en ligne

Virtualization

CPU
13 VM 33%
pxm-08 en ligne

Virtualization

CPU
10 VM 21%
pxm-09 en ligne

Virtualization

CPU
15 VM 39%
pxm-10 en ligne

Virtualization

CPU
8 VM 18%
pxm-11 en ligne

Virtualization

CPU
12 VM 25%
pxm-12 en ligne

Virtualization

CPU
17 VM 44%
pxm-13 en ligne

Virtualization

CPU
7 VM 19%
pxm-14 en ligne

Virtualization

CPU
10 VM 23%
pxm-15 en ligne

Virtualization

CPU
9 VM 20%
pxm-16 en ligne

Virtualization

CPU
3 VM 27%
pxm-17 en ligne

Virtualization

CPU
2 VM 30%
pxm-18 en ligne

Virtualization

CPU
1 VM 16%
pxm-19 en ligne

Virtualization

CPU
3 VM 35%
pxm-20 en ligne

Virtualization

CPU
0 VM 14%
pxm-21 en ligne

Virtualization

CPU
2 VM 29%
pxm-22 en ligne

Virtualization

CPU
1 VM 22%
pxm-23 en ligne

Virtualization

CPU
3 VM 26%
pxm-24 en ligne

Virtualization

CPU
0 VM 17%
pxm-25 en ligne

Virtualization

CPU
2 VM 24%

Rocks Cluster

8 en ligne
rocks-01 Head node
rocks-02 Compute node
rocks-03 Compute node
rocks-04 Compute node
rocks-05 Compute node
rocks-06 Compute node
rocks-07 Compute node
rocks-08 Compute node

Inférence IA

10 en ligne
inf-01 AI inference
inf-02 Model API
inf-03 Computer vision
inf-04 NLP inference
inf-05 Batch inference
inf-06 Student workloads
inf-07 Demo services
inf-08 GPU services
inf-09 Data services
inf-10 Reserve inference