ARC users must abide by the terms and conditions stipulated by the University of Oxford as a condition for using IT systems and the computer usage rules and etiquette laid out by IT Services.
System security is vital to protect user activities and maintain a reliable service. The ARC team reserve the right to refuse any installation of applications, underpinning software or services in its computing environments that, in the sole considered opinion of its experts, is likely to compromise the integrity, availability or security of those environments or is likely to impact the service for other users.
Also, the ARC team may de-activate any account without notice, where the user of the account has installed software that is disruptive or designed to circumvent security without the knowledge of the ARC team.
The login nodes can be exclusively used for preparing, launching and monitoring batch jobs. Such preparation includes copying and uncompressing files, compiling source code and building applications and editing text files.
No compute intensive application should be run on the login nodes for an extended period of time. All production processing should be carried out on the compute nodes, making use of the batch system. Batch processing is, by its nature, non-interactive and should, ideally, also be executed in parallel, i.e. through the use of the multiple cores available on a compute node or (depending on service) through the use of several compute nodes at a time.
Batch jobs are limited to a walltime of 120 hours for the standard service. Many scientific applications have the capability of checkpointing at regular prescribed intervals, with the ability of resuming processing by restarting from a checkpoint. Checkpointing is the preferred way of ensuring the walltime does not exceed the limit imposed, and users are strongly encouraged to use checkpointing where this capability exists. However, email please contact us if there is no alternative but to extend walltime beyond the limit for certain jobs.
Each user of the ARC service is allocated a disk quota, see details given in the Storage help page.
While every effort is made to ensure the integrity of data stored by users on the service storage, ARC is under no obligation to guarantee the integrity or availability of data. Thus, the ARC team does not accept any liability, financial or otherwise for loss of data. The guarantee of data integrity and availablity is the responsibility the individual users, who should employ standard industry practice for their important data and store it at sites other than ARC, for example, in their department.
The ARC storage system is optimised for high performance computing and is thus more expensive than systems designed for long-term data storage or archiving. ARC users should therefore use ARC storage exclusively for the use of live research data. ARC does not back data up, nor does it archive data or store it long term. The ARC team reserves the right to purge data that has not been used for a long time from storage, in order to make space for new users and maintain system performance. In the event that ARC plans to purge data, affected users are notified to allow them to find alternative storage.
Users can read / write their own data - unless they set Linux permissions to allow other users access. The ARC systems' filesystems are not encrypted and users must discuss storage of sensitive data with the ARC team prior to placing it on our systems. Normal good IT practice is expected e.g. do not share account passwords.
Users of the ARC service must acknowledge the use of ARC in any publication that contains results obtained using the ARC resources. Such a publication must include the following text, or citation, including the DOI:
The authors would like to acknowledge the use of the University of Oxford Advanced Research Computing (ARC) facility in carrying out this work. http://dx.doi.org/10.5281/zenodo.22558
Alternatively, if it is not possible to include the above, the ARC facility may be cited as:
Richards, Andrew. (2015). University of Oxford Advanced Research Computing. Zenodo.10.5281/zenodo.22558
The ARC facility maintains a bibliographic database of all publications that have benefited from use of its resources. Members must notify the ARC team of all appropriate publications, by emailing the full bibliographic references to firstname.lastname@example.org .