-
Notifications
You must be signed in to change notification settings - Fork 0
6. External access for the HPC
- ""
- ""
- ""
To ensure robust security measures, the High-Performance Computing (HPC) system maintains limited connectivity beyond the local network of the Norwich Research Park's (NRP) campus. While this setup provides a secure environment, it can present challenges when accessing public data from external sources. There are several options for users to overcome this, depending on their individual needs.
One approach involves downloading data to a designated project area by mounting the relevant project directory locally, and executing the download via that local system. While this method is feasible for moderate-sized datasets, it can be impractical for larger files. In such cases, users can request assistance, and EI scientific computing support (Martin Ayling) who will retrieve the data on their behalf, storing it securely in the /ei/public/ directory if appropriate. Commonly accessed databases such as nr and nt are regularly updated to ensure users have access to the latest versions.
For smaller file transfers (up to 30GB), researchers can use the NBI drop-off service (https://dropoff.nbi.ac.uk/). This platform provides a convenient and secure means of transferring files to and from external collaborators using a browser, which is very intuitive.
For larger transfers with collaborators, the NRP runs a Globus instance. With Globus, users can seamlessly move large datasets between the HPC system and external storage systems, collaborators, or research institutions. Instructions detailing the setup and operation of globus are covered in a separate episode.
- Induction
- HPC Best practice
- Job Arrays - RC documentation
- Methods to Improve I/O Performance - RC documentation
- Customising your bash profile for ease and efficiency
- Customise bash profile: Logging Your Command History Automatically
- Using the ei-gpu partition on the Earlham Institute computing cluster
- Using the GPUs at EI
- HPC Job Summary Tool
- EI Cloud (CyVerse)
- Git and GitHub
- Worked examples
- Job Arrays
- Using Parabricks on the GPUs
- dependencies
- Software installations
- Workflow management system
- Transfers
- Local (mounting HPC storage)
- Remote - <1gb (ood)
- Remote - <50gb (nbi drop off)
- Remote - No limit (globus)
- mv command