# Manage an AI server
Source: https://docs.infraista.com/en/compute/ai-servers/manage-an-ai-server



An AI server's page has the same tabs and buttons as an [instance's](/en/compute/instances/your-server-page),
so the instance guides apply as they are:

| To                                               | See                                                                      |
| ------------------------------------------------ | ------------------------------------------------------------------------ |
| Start, stop or restart it                        | [Start, stop and restart](/en/compute/instances/start-stop-and-restart)  |
| Set firewall rules, reverse DNS or a floating IP | [Firewall and networking](/en/compute/instances/firewall-and-networking) |
| Back it up or restore it                         | [Back up and restore](/en/compute/instances/backups)                     |
| Switch to monthly billing                        | [Monthly billing](/en/compute/instances/monthly-billing)                 |
| Rename, move or delete it                        | [Rename, move or delete](/en/compute/instances/rename-move-or-delete)    |

## What's different on AI servers [#whats-different-on-ai-servers]

* **GPU metrics**: the **Metrics** tab adds a **GPU** graph with utilization and VRAM in use, with the
  GPU model in its title. See [Metrics and activity](/en/compute/instances/metrics-and-activity).
* **Resizing** offers only plans with the same GPU model. Disks can grow but never shrink. See
  [Resize a server](/en/compute/instances/resize-a-server).
* **Stopping saves money**: hourly AI servers aren't charged for compute while they're stopped. Stop
  them between render jobs or training runs.

## Related [#related]

* [Choose a GPU package](/en/compute/ai-servers/choose-a-gpu-package)