Nexus/Network: Difference between revisions
Jump to navigation
Jump to search
No edit summary |
No edit summary |
||
| Line 1: | Line 1: | ||
==Overview== | ==Overview== | ||
The [[Nexus]] cluster runs on a [https://en.wikipedia.org/wiki/Hierarchical_internetworking_model hierarchical] [https://en.wikipedia.org/wiki/Ethernet Ethernet]-based network with node-level speeds ranging anywhere from [https://en.wikipedia.org/wiki/Gigabit_Ethernet 1GbE] to [https://en.wikipedia.org/wiki/100_Gigabit_Ethernet 100GbE]. Generally speaking, newer-purchased nodes often come with hardware capable of using faster speeds, but not always. Faster speeds require increasingly more powerful/expensive network switches and cables, so some labs/centers have opted to stay with slower speeds. | The [[Nexus]] cluster runs on a [https://en.wikipedia.org/wiki/Hierarchical_internetworking_model hierarchical] [https://en.wikipedia.org/wiki/Ethernet Ethernet]-based network with node-level speeds ranging anywhere from [https://en.wikipedia.org/wiki/Gigabit_Ethernet 1GbE] to [https://en.wikipedia.org/wiki/100_Gigabit_Ethernet 100GbE]. Generally speaking, newer-purchased nodes often come with hardware capable of using, and therefore use, faster speeds, but not always. Faster speeds require increasingly more powerful/expensive network switches and cables, so some labs/centers have opted to stay with slower speeds. | ||
In the future, SLURM's [https://slurm.schedmd.com/topology.html topology-aware resource allocation support] may be implemented on the cluster, but it is not currently. | In the future, SLURM's [https://slurm.schedmd.com/topology.html topology-aware resource allocation support] may be implemented on the cluster, but it is not currently. | ||
Revision as of 19:05, 26 November 2024
Overview
The Nexus cluster runs on a hierarchical Ethernet-based network with node-level speeds ranging anywhere from 1GbE to 100GbE. Generally speaking, newer-purchased nodes often come with hardware capable of using, and therefore use, faster speeds, but not always. Faster speeds require increasingly more powerful/expensive network switches and cables, so some labs/centers have opted to stay with slower speeds.
In the future, SLURM's topology-aware resource allocation support may be implemented on the cluster, but it is not currently.