About Scicluster
A Gnu/Linux Cluster - one machine, consisting of many machines
On one hand you can look at Gnu/Linux clusters as rather large and powerful supercomputers, but on the other hand you can look at them as just a large bunch of servers and some storage system(s) connected with each other through a (high speed) network. Both of these views are fully correct, and it's therefore important to be aware of the strengths and the limitations of such a system.
Rocks Gnu/Linux operating system
Scicluster is build upon Rocks version 7 cluster distribution. Rocks is an open-source Gnu/Linux cluster distribution based upon CentOS 7.4 that enables end users to easily build computational clusters. Hundreds of researchers from around the world have used Rocks to deploy their own clusters.
Resource description
Key info about any cluster including Scicluster are compute nodes, interconnection network, operating system and storage configuration.
Node info
| Name | CPU/GPU | architecture | # cores | RAM (GB) | Model |
| sci (the head/login node) | 2 x Xeon E5-2420 | Ivy Bridge EN (2012) | 2 x 6 | 16 | Supermicro |
| compute-0-0 | 2 x Xeon E5-2630 Quadro P4000 |
Broadwell (2014) Pascal (2016) |
2 x 10 1792 |
64 8 |
ML350-Gen9 |
| compute-0-1 | 2 x Xeon E5-2695 | Broadwell (2014) | 2 x 18 | 64 | DL360-Gen9 |
| compute-0-2 | 2 x Xeon E5-2690 | Haswell (2013) | 2 x 12 | 256 | DL380-Gen9 |
| compute-0-3 | 2 x Xeon E5-2695 Tesla K80 |
Broadwell (2014) Kepler (2014) |
2 x 18 4992 |
64 24 |
DL380-Gen9 |
Networks
Currently all nodes are connected with a 1 Gb ethernet network for IO/management as well as a 10 Gb low latency one for message passing (see the above figure).
OS
CentOS 7.4
Storage
- Scratch space ~ 4.8 TB local high speed disk (1.2 TB on each compute node)
- Shared space ~ 24 TB network file system (NFS) installed on the head node
For more details see Storage section.