Computing resources

Which resources can you use?

You must not transfer data onto personal devices or third party cloud services, unless it is publicly available, open source or you have appropriate permission.

Laptop

Your laptop will be supplied by the University. We recommend requesting a Windows or Max device, as this allows you to use desktop versions of the Office 365 apps (especially Teams – which we use daily).

We recommend using your local machine for small projects or those where you’d like to use other software locally without context switching. For development we recommend using a Unix-like environment. On Windows, we recommend using Windows Subsystem for Linux (WSL).

Important

Always make sure your data is backed up. OneDrive syncing should be set up automatically on your device, although for users of WSL this requires additional set-up. 1

For most documents, the appropriate location on SharePoint is the best place to store things. You can add SharePoint shortcuts to your OneDrive, meaning those folders get synced too. Or you can sync entire SharePoint libraries, but this is often too much for the OneDrive sync client to handle!

JGI server

We run our own server (hostname: IT160062.users.bris.ac.uk) running Ubuntu. It has:

  • 12 cores (24 threads),
  • 256 GB of memory,
  • 10 GB NVIDIA graphics memory (for deep learning experiments),
  • 4 TB of data storage.

Resources are shared between all JGI data scientists on an informal basis.

You can get SSH access from one of the other data scientists in the team.

High Performance Computing (HPC) clusters

The University runs the BlueCrystal 4 and BluePebble HPC clusters.

We have a JGI project (project code: JGIN025022) that you ask to join when you apply for access to the University HPC.

Once granted access, you will need to be within campus or use the University VPN to connect to the HPC login servers before you can submit jobs.

There are also national HPC facilities for specific uses, for example JASMIN for environmental science.

Important

The HPC systems are not backed up and you need to transfer important files to GitHub or the RDSF.

Research Data Storage Facility (RDSF)

The RDSF provides long-term data storage for research projects. Each user gets 5 TB or stage for free.

If you are a project’s principal investigator, you can apply for RDSF storage by becoming a data steward for your project.

We have a JGI project (name: JGI_Training_Project) that we use for familiarisation and training purposes. You can get access from one of the other data scientists in the team.

You will need to be within campus or use the University VPN to connect to the RDSF. Depending on your project’s settings, this storage may also be accessible from HPC and/or the JGI server.

Other servers

Some project and/or groups have access to their own server (like the JGI server) or custom VMs.

Depending on the specific requirements, these could be used to host:

  • a Trusted Research Environment (TRE), commonly also called a data safe haven or secure research service (SRS),
  • a externally-facing web server to provide data ingress/egress.

Footnotes

  1. There is an unofficial OneDrive client for Linux that works under WSL. This can be used to selectively sync a WSL directory to OneDrive, for backup purposes.↩︎