Overview¶
What is the TML Cluster?¶
The TML cluster is a set of computers connected together to run heavy simulations. Instead of running a simulation on your own PC, you submit it to the cluster, which has much more computing power.
You access the cluster remotely from your PC using SSH (Secure Shell) — a program that lets you control a remote computer through a terminal.
How It Works¶
- You connect to the head node from your PC via SSH
- You submit your simulation job on the head node
- SLURM (Simple Linux Utility for Resource Management), the job scheduler, assigns your job to available computing nodes
- The computing nodes run the calculation
- Results are saved to the NAS (Network Attached Storage), which you can access from the head node
The cluster only allows inbound connections from specific IP addresses. Each lab room has a router so that all computers in the room share a single gateway IP that is whitelisted by the cluster.
Components¶
| Component | Role |
|---|---|
| Head node | The computer you log into. Used for submitting jobs and managing files. Does not run calculations. |
| Computing nodes (t01~t14) | 14 computers that run simulations (t03, t04, t05, t14 are currently powered off). You don't log into these directly — SLURM handles them. |
| NAS | Shared storage on the network. Your home directory and data files live here. All nodes can access it. |
| Network switch | Connects all nodes and storage together. |
| Router | Installed in each lab room. Provides a single gateway IP that is whitelisted by the cluster. |
Terminology¶
| Term | Meaning |
|---|---|
| SSH (Secure Shell) | A program to securely log into a remote computer via terminal. This is how you connect to the cluster. |
| Node | A computer in the cluster. |
| Head node | The node you log into. Also called "master node." Handles job submission and file management. |
| Computing node | A node that runs calculations. Managed by SLURM. |
| NAS (Network Attached Storage) | Shared hard drive on the network. All nodes read and write files here. |
| Server | A computer that provides a service (file storage, remote access, etc.) |
| Client | Your PC, when it's connecting to a server. |
| Hostname | A name for a computer (e.g. t01, t02), used instead of its IP address. |
| IP (Internet Protocol) address | A numeric address for a computer, e.g. 192.168.0.1. |
| Job | A simulation or calculation task submitted to the cluster. |
| SLURM (Simple Linux Utility for Resource Management) | The scheduler that decides which computing node runs your job and when. |