After more than 9 months of development. Sie has officially launched its new Operating System for Cluster HPC and scientific computing machines. Said software is based on Centos 6.3 with Epel repositories, both on server and nodes. Having the same system throughout the system allows us a perfect inter-node integration and facilities at the time of installations.
Scalability in HPC cluster
As a queue manager, Open Grid Scheduler has been installed, due to the increasing number of cores per machine, it guarantees correct scaling and parallelism with low-latency network systems such as 10g or Infiniband.
The node installation system has been greatly improved. The kickstarted PXE system installs nodes over the network in a matter of minutes. Subsequently, a unique post installation script for each system completes the installation and integration. Once installed, the nodes work with their own hard drive, freeing up network load.
Our new system is compatible with practically all calculation tools. It has Kernell el6, based on Red Hat. Thanks to this, it is possible to install InfiniBand mlnx proprietary drivers or “enterprise” type applications.
The new services of Ganglia, Apcupsd (compatible SAIs) and various web tools have been modified that allow cluster control to be more friendly and simple.
SIE has launched a cluster recycling and updating campaign. If you are interested in learning more about the new system, contact us!
