Hydra cluster

Hydra is defined as a balanced cluster, i.e. suitable for batch, multi-cores (SMP) and multi-nodes (MPI) jobs. The current cluster is composed of a total of 153 nodes hosting ~1200 cores with 32 to 256 GB of RAM per node and interconnected with 1 Gbps ethernet and 10 Gbps/40 Gbps Infiniband networks. The total disk space available to store the jobs output (location of your work directory) is ~20 TB.
There are 4 master nodes hosting all the cluster services such as the jobs manager and scheduler, the LDAP server and the NFS exports to the compute nodes.
See the detailed description of the Hydra installation.


There are two sources of documentation for Hydra:
HowTos: a set of pages describing how to perform common operations on the Hydra cluster.
    Topics are: Q&A: questions and answers.


Software or library installation

Service to access or request software, package or library installation. After an installation, the programs/libraries will be automatically accessible to all Hydra users. You can ask to install you own applications.