Skip to content

Storage and Data

Whether its datasets, input files, output files, or scripts, jobs typically require data to run. This page highlights the different locations where data can be stored, how each are used, and how you can access them.

Backing Up Data

The Blugold Center for High-Performance Computing does not currently back up any user, group, or class data on our clusters. To prevent any potential data loss, we recommend frequently backing up your most important files to locations off the cluster. See our file transfers guide on different ways you can copy your files.

User Data

Also known as your 'home directory', user data is where you can find your own personal files that typically only you have access to. This is the default starting directory for most users when the log into the cluster.

Location: /data/users/<username>

Instead of typing the full path, in many cases you can use the $HOME environment variable or ~ to refer to your home directory. Note that not all scripts will support one or both of these aliases.

[myuser@bose ~]$ echo $HOME
/data/users/myuser

[myuser@bose somedirectory]$ cd ~/myfolder/      # Change directory starting from home directory "~"
[myuser@bose myfolder]$ pwd         # Print working directory
/data/users/myuser/myfolder

Group Data

Every research group is able to have space provisioned under /data/groups where faculty and students can share files, scripts, and datasets. This is highly recommended to keep everything under one location on a per-project basis and ensure continued access after group members change or graduate.

More about research group storage ».

Class Data

All classes that use the cluster for any purpose are provisioned space under /data/classes when given access by the HPC Team. By default, each student in the class is given their own folder to work out of, but project group folders can be created as well with custom access per-student.

More about class storage ».

Pre-Installed Datasets

The HPC Team maintains a repository of datasets and models available for some software to limit duplication of large files. If you have recommendations on what we should include, please contact the HPC team.

Software: